Statistics
| Branch: | Tag: | Revision:

root / NEWS @ 5349519d

History | View | Annotate | Download (130.1 kB)

1
News
2
====
3

    
4

    
5
Version 2.12.0 alpha1
6
---------------------
7

    
8
*(unreleased)*
9

    
10
Incompatible/important changes
11
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
12

    
13
- Do not use debug mode in production. Certain daemons will issue warnings
14
  when launched in debug mode. Some debug logging violates some of the new
15
  invariants in the system (see "New features"). The logging has been kept as
16
  it aids diagnostics and development.
17

    
18
New features
19
~~~~~~~~~~~~
20

    
21
- OS install script parameters now come in public, private and secret
22
  varieties:
23

    
24
  - Public parameters are like all other parameters in Ganeti.
25
  - Ganeti will not log private and secret parameters, *unless* it is running
26
    in debug mode.
27
  - Ganeti will not save secret parameters to configuration. Secret parameters
28
    must be supplied every time you install, or reinstall, an instance.
29
  - Attempting to override public parameters with private or secret parameters
30
    results in an error. Similarly, you may not use secret parameters to
31
    override private parameters.
32

    
33

    
34
Version 2.11.0 alpha1
35
---------------------
36

    
37
*(unreleased)*
38

    
39
Incompatible/important changes
40
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
41

    
42
- ``gnt-node list`` no longer shows disk space information for shared file
43
  disk templates because it is not a node attribute. (For example, if you have
44
  both the file and shared file disk templates enabled, ``gnt-node list`` now
45
  only shows information about the file disk template.)
46
- The shared file disk template is now in the new 'sharedfile' storage type.
47
  As a result, ``gnt-node list-storage -t file`` now only shows information
48
  about the file disk template and you may use ``gnt-node list-storage -t
49
  sharedfile`` to query storage information for the shared file disk template.
50
- Over luxi, syntactially incorrect queries are now rejected as a whole;
51
  before, a 'SumbmitManyJobs' request was partially executed, if the outer
52
  structure of the request was syntactically correct. As the luxi protocol
53
  is internal (external applications are expected to use RAPI), the impact
54
  of this incompatible change should be limited.
55
- Queries for nodes, instances, groups, backups and networks are now
56
  exclusively done via the luxi daemon. Legacy python code was removed,
57
  as well as the --enable-split-queries configuration option.
58
- Orphan volumes errors are demoted to warnings and no longer affect the exit
59
  code of ``gnt-cluster verify``.
60
- RPC security got enhanced by using different client SSL certificates
61
  for each node. In this context 'gnt-cluster renew-crypto' got a new
62
  option '--renew-node-certificates', which renews the client
63
  certificates of all nodes. After a cluster upgrade from pre-2.11, run
64
  this to create client certificates and activate this feature.
65

    
66
New features
67
~~~~~~~~~~~~
68

    
69
- Instance moves, backups and imports can now use compression to transfer the
70
  instance data.
71
- Node groups can be configured to use an SSH port different than the
72
  default 22.
73
- Added experimental support for Gluster distributed file storage as the
74
  ``gluster`` disk template under the new ``sharedfile`` storage type through
75
  automatic management of per-node FUSE mount points. You can configure the
76
  mount point location at ``gnt-cluster init`` time by using the new
77
  ``--gluster-storage-dir`` switch.
78
- Job scheduling is now handled by luxid, and the maximal number of jobs running
79
  in parallel is a run-time parameter of the cluster.
80

    
81
New dependencies
82
~~~~~~~~~~~~~~~~
83
The following new dependencies have been added:
84

    
85
For Haskell:
86

    
87
- ``zlib`` library (http://hackage.haskell.org/package/base64-bytestring)
88

    
89
- ``base64-bytestring`` library (http://hackage.haskell.org/package/zlib),
90
  at least version 1.0.0.0
91

    
92
- ``lifted-base`` library (http://hackage.haskell.org/package/lifted-base)
93

    
94

    
95
Version 2.10.0
96
--------------
97

    
98
*(Released Thu, 20 Feb 2014)*
99

    
100
Incompatible/important changes
101
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
102

    
103
- Adding disks with 'gnt-instance modify' now waits for the disks to sync per
104
  default. Specify --no-wait-for-sync to override this behavior.
105
- The Ganeti python code now adheres to a private-module layout. In particular,
106
  the module 'ganeti' is no longer in the python search path.
107
- On instance allocation, the iallocator now considers non-LVM storage
108
  properly. In particular, actual file storage space information is used
109
  when allocating space for a file/sharedfile instance.
110
- When disabling disk templates cluster-wide, the cluster now first
111
  checks whether there are instances still using those templates.
112
- 'gnt-node list-storage' now also reports storage information about
113
  file-based storage types.
114
- In case of non drbd instances, export \*_SECONDARY environment variables
115
  as empty strings (and not "None") during 'instance-migrate' related hooks.
116

    
117
New features
118
~~~~~~~~~~~~
119

    
120
- KVM hypervisors can now access RBD storage directly without having to
121
  go through a block device.
122
- A new command 'gnt-cluster upgrade' was added that automates the upgrade
123
  procedure between two Ganeti versions that are both 2.10 or higher.
124
- The move-instance command can now change disk templates when moving
125
  instances, and does not require any node placement options to be
126
  specified if the destination cluster has a default iallocator.
127
- Users can now change the soundhw and cpuid settings for XEN hypervisors.
128
- Hail and hbal now have the (optional) capability of accessing average CPU
129
  load information through the monitoring deamon, and to use it to dynamically
130
  adapt the allocation of instances.
131
- Hotplug support. Introduce new option '--hotplug' to ``gnt-instance modify``
132
  so that disk and NIC modifications take effect without the need of actual
133
  reboot. There are a couple of constrains currently for this feature:
134

    
135
   - only KVM hypervisor (versions >= 1.0) supports it,
136
   - one can not (yet) hotplug a disk using userspace access mode for RBD
137
   - in case of a downgrade instances should suffer a reboot in order to
138
     be migratable (due to core change of runtime files)
139

    
140
Misc changes
141
~~~~~~~~~~~~
142

    
143
- A new test framework for logical units was introduced and the test
144
  coverage for logical units was improved significantly.
145
- Opcodes are entirely generated from Haskell using the tool 'hs2py' and
146
  the module 'src/Ganeti/OpCodes.hs'.
147
- Constants are also generated from Haskell using the tool
148
  'hs2py-constants' and the module 'src/Ganeti/Constants.hs', with the
149
  exception of socket related constants, which require changing the
150
  cluster configuration file, and HVS related constants, because they
151
  are part of a port of instance queries to Haskell.  As a result, these
152
  changes will be part of the next release of Ganeti.
153

    
154
New dependencies
155
~~~~~~~~~~~~~~~~
156

    
157
The following new dependencies have been added/updated.
158

    
159
Python
160

    
161
- The version requirements for ``python-mock`` have increased to at least
162
  version 1.0.1. It is still used for testing only.
163

    
164
Since 2.10.0 rc3
165
~~~~~~~~~~~~~~~~
166

    
167
- Fix integer overflow problem in hbal
168

    
169

    
170
Version 2.10.0 rc3
171
------------------
172

    
173
*(Released Wed, 12 Feb 2014)*
174

    
175
This was the third RC release of the 2.10 series. Since 2.10.0 rc2:
176

    
177
- Improved hotplug robustness
178
- Start Ganeti daemons after ensure-dirs during upgrade
179
- Documentation improvements
180

    
181
Inherited from the 2.9 branch:
182

    
183
- Fix the RAPI instances-multi-alloc call
184
- assign unique filenames to file-based disks
185
- gracefully handle degraded non-diskless instances with 0 disks (issue 697)
186
- noded now runs with its specified group, which is the default group,
187
  defaulting to root (issue 707)
188
- make using UUIDs to identify nodes in gnt-node consistently possible
189
  (issue 703)
190

    
191

    
192
Version 2.10.0 rc2
193
------------------
194

    
195
*(Released Fri, 31 Jan 2014)*
196

    
197
This was the second RC release of the 2.10 series. Since 2.10.0 rc1:
198

    
199
- Documentation improvements
200
- Run drbdsetup syncer only on network attach
201
- Include target node in hooks nodes for migration
202
- Fix configure dirs
203
- Support post-upgrade hooks during cluster upgrades
204

    
205
Inherited from the 2.9 branch:
206

    
207
- Ensure that all the hypervisors exist in the config file (Issue 640)
208
- Correctly recognise the role as master node (Issue 687)
209
- configure: allow detection of Sphinx 1.2+ (Issue 502)
210
- gnt-instance now honors the KVM path correctly (Issue 691)
211

    
212
Inherited from the 2.8 branch:
213

    
214
- Change the list separator for the usb_devices parameter from comma to space.
215
  Commas could not work because they are already the hypervisor option
216
  separator (Issue 649)
217
- Add support for blktap2 file-driver (Issue 638)
218
- Add network tag definitions to the haskell codebase (Issue 641)
219
- Fix RAPI network tag handling
220
- Add the network tags to the tags searched by gnt-cluster search-tags
221
- Fix caching bug preventing jobs from being cancelled
222
- Start-master/stop-master was always failing if ConfD was disabled. (Issue 685)
223

    
224

    
225
Version 2.10.0 rc1
226
------------------
227

    
228
*(Released Tue, 17 Dec 2013)*
229

    
230
This was the first RC release of the 2.10 series. Since 2.10.0 beta1:
231

    
232
- All known issues in 2.10.0 beta1 have been resolved (see changes from
233
  the 2.8 branch).
234
- Improve handling of KVM runtime files from earlier Ganeti versions
235
- Documentation fixes
236

    
237
Inherited from the 2.9 branch:
238

    
239
- use custom KVM path if set for version checking
240
- SingleNotifyPipeCondition: don't share pollers
241

    
242
Inherited from the 2.8 branch:
243

    
244
- Fixed Luxi daemon socket permissions after master-failover
245
- Improve IP version detection code directly checking for colons rather than
246
  passing the family from the cluster object
247
- Fix NODE/NODE_RES locking in LUInstanceCreate by not acquiring NODE_RES locks
248
  opportunistically anymore (Issue 622)
249
- Allow link local IPv6 gateways (Issue 624)
250
- Fix error printing (Issue 616)
251
- Fix a bug in InstanceSetParams concerning names: in case no name is passed in
252
  disk modifications, keep the old one. If name=none then set disk name to
253
  None.
254
- Update build_chroot script to work with the latest hackage packages
255
- Add a packet number limit to "fping" in master-ip-setup (Issue 630)
256
- Fix evacuation out of drained node (Issue 615)
257
- Add default file_driver if missing (Issue 571)
258
- Fix job error message after unclean master shutdown (Issue 618)
259
- Lock group(s) when creating instances (Issue 621)
260
- SetDiskID() before accepting an instance (Issue 633)
261
- Allow the ext template disks to receive arbitrary parameters, both at creation
262
  time and while being modified
263
- Xen handle domain shutdown (future proofing cherry-pick)
264
- Refactor reading live data in htools (future proofing cherry-pick)
265

    
266

    
267
Version 2.10.0 beta1
268
--------------------
269

    
270
*(Released Wed, 27 Nov 2013)*
271

    
272
This was the first beta release of the 2.10 series. All important changes
273
are listed in the latest 2.10 entry.
274

    
275
Known issues
276
~~~~~~~~~~~~
277

    
278
The following issues are known to be present in the beta and will be fixed
279
before rc1.
280

    
281
- Issue 477: Wrong permissions for confd LUXI socket
282
- Issue 621: Instance related opcodes do not aquire network/group locks
283
- Issue 622: Assertion Error: Node locks differ from node resource locks
284
- Issue 623: IPv6 Masterd <-> Luxid communication error
285

    
286

    
287
Version 2.9.4
288
-------------
289

    
290
*(Released Mon, 10 Feb 2014)*
291

    
292
- Fix the RAPI instances-multi-alloc call
293
- assign unique filenames to file-based disks
294
- gracefully handle degraded non-diskless instances with 0 disks (issue 697)
295
- noded now runs with its specified group, which is the default group,
296
  defaulting to root (issue 707)
297
- make using UUIDs to identify nodes in gnt-node consistently possible
298
  (issue 703)
299

    
300

    
301
Version 2.9.3
302
-------------
303

    
304
*(Released Mon, 27 Jan 2014)*
305

    
306
- Ensure that all the hypervisors exist in the config file (Issue 640)
307
- Correctly recognise the role as master node (Issue 687)
308
- configure: allow detection of Sphinx 1.2+ (Issue 502)
309
- gnt-instance now honors the KVM path correctly (Issue 691)
310

    
311
Inherited from the 2.8 branch:
312

    
313
- Change the list separator for the usb_devices parameter from comma to space.
314
  Commas could not work because they are already the hypervisor option
315
  separator (Issue 649)
316
- Add support for blktap2 file-driver (Issue 638)
317
- Add network tag definitions to the haskell codebase (Issue 641)
318
- Fix RAPI network tag handling
319
- Add the network tags to the tags searched by gnt-cluster search-tags
320
- Fix caching bug preventing jobs from being cancelled
321
- Start-master/stop-master was always failing if ConfD was disabled. (Issue 685)
322

    
323

    
324
Version 2.9.2
325
-------------
326

    
327
*(Released Fri, 13 Dec 2013)*
328

    
329
- use custom KVM path if set for version checking
330
- SingleNotifyPipeCondition: don't share pollers
331

    
332
Inherited from the 2.8 branch:
333

    
334
- Fixed Luxi daemon socket permissions after master-failover
335
- Improve IP version detection code directly checking for colons rather than
336
  passing the family from the cluster object
337
- Fix NODE/NODE_RES locking in LUInstanceCreate by not acquiring NODE_RES locks
338
  opportunistically anymore (Issue 622)
339
- Allow link local IPv6 gateways (Issue 624)
340
- Fix error printing (Issue 616)
341
- Fix a bug in InstanceSetParams concerning names: in case no name is passed in
342
  disk modifications, keep the old one. If name=none then set disk name to
343
  None.
344
- Update build_chroot script to work with the latest hackage packages
345
- Add a packet number limit to "fping" in master-ip-setup (Issue 630)
346
- Fix evacuation out of drained node (Issue 615)
347
- Add default file_driver if missing (Issue 571)
348
- Fix job error message after unclean master shutdown (Issue 618)
349
- Lock group(s) when creating instances (Issue 621)
350
- SetDiskID() before accepting an instance (Issue 633)
351
- Allow the ext template disks to receive arbitrary parameters, both at creation
352
  time and while being modified
353
- Xen handle domain shutdown (future proofing cherry-pick)
354
- Refactor reading live data in htools (future proofing cherry-pick)
355

    
356

    
357
Version 2.9.1
358
-------------
359

    
360
*(Released Wed, 13 Nov 2013)*
361

    
362
- fix bug, that kept nodes offline when readding
363
- when verifying DRBD versions, ignore unavailable nodes
364
- fix bug that made the console unavailable on kvm in split-user
365
  setup (issue 608)
366
- DRBD: ensure peers are UpToDate for dual-primary (inherited 2.8.2)
367

    
368

    
369
Version 2.9.0
370
-------------
371

    
372
*(Released Tue, 5 Nov 2013)*
373

    
374
Incompatible/important changes
375
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
376

    
377
- hroller now also plans for capacity to move non-redundant instances off
378
  any node to be rebooted; the old behavior of completely ignoring any
379
  non-redundant instances can be restored by adding the --ignore-non-redundant
380
  option.
381
- The cluster option '--no-lvm-storage' was removed in favor of the new option
382
  '--enabled-disk-templates'.
383
- On instance creation, disk templates no longer need to be specified
384
  with '-t'. The default disk template will be taken from the list of
385
  enabled disk templates.
386
- The monitoring daemon is now running as root, in order to be able to collect
387
  information only available to root (such as the state of Xen instances).
388
- The ConfD client is now IPv6 compatible.
389
- File and shared file storage is no longer dis/enabled at configure time,
390
  but using the option '--enabled-disk-templates' at cluster initialization and
391
  modification.
392
- The default directories for file and shared file storage are not anymore
393
  specified at configure time, but taken from the cluster's configuration.
394
  They can be set at cluster initialization and modification with
395
  '--file-storage-dir' and '--shared-file-storage-dir'.
396
- Cluster verification now includes stricter checks regarding the
397
  default file and shared file storage directories. It now checks that
398
  the directories are explicitely allowed in the 'file-storage-paths' file and
399
  that the directories exist on all nodes.
400
- The list of allowed disk templates in the instance policy and the list
401
  of cluster-wide enabled disk templates is now checked for consistency
402
  on cluster or group modification. On cluster initialization, the ipolicy
403
  disk templates are ensured to be a subset of the cluster-wide enabled
404
  disk templates.
405

    
406
New features
407
~~~~~~~~~~~~
408

    
409
- DRBD 8.4 support. Depending on the installed DRBD version, Ganeti now uses
410
  the correct command syntax. It is possible to use different DRBD versions
411
  on different nodes as long as they are compatible to each other. This
412
  enables rolling upgrades of DRBD with no downtime. As permanent operation
413
  of different DRBD versions within a node group is discouraged,
414
  ``gnt-cluster verify`` will emit a warning if it detects such a situation.
415
- New "inst-status-xen" data collector for the monitoring daemon, providing
416
  information about the state of the xen instances on the nodes.
417
- New "lv" data collector for the monitoring daemon, collecting data about the
418
  logical volumes on the nodes, and pairing them with the name of the instances
419
  they belong to.
420
- New "diskstats" data collector, collecting the data from /proc/diskstats and
421
  presenting them over the monitoring daemon interface.
422
- The ConfD client is now IPv6 compatible.
423

    
424
New dependencies
425
~~~~~~~~~~~~~~~~
426
The following new dependencies have been added.
427

    
428
Python
429

    
430
- ``python-mock`` (http://www.voidspace.org.uk/python/mock/) is now a required
431
  for the unit tests (and only used for testing).
432

    
433
Haskell
434

    
435
- ``hslogger`` (http://software.complete.org/hslogger) is now always
436
  required, even if confd is not enabled.
437

    
438
Since 2.9.0 rc3
439
~~~~~~~~~~~~~~~
440

    
441
- Correctly start/stop luxid during gnt-cluster master-failover (inherited
442
  from stable-2.8)
443
- Improved error messsages (inherited from stable-2.8)
444

    
445

    
446
Version 2.9.0 rc3
447
-----------------
448

    
449
*(Released Tue, 15 Oct 2013)*
450

    
451
The third release candidate in the 2.9 series. Since 2.9.0 rc2:
452

    
453
- in implicit configuration upgrade, match ipolicy with enabled disk templates
454
- improved harep documentation (inherited from stable-2.8)
455

    
456

    
457
Version 2.9.0 rc2
458
-----------------
459

    
460
*(Released Wed, 9 Oct 2013)*
461

    
462
The second release candidate in the 2.9 series. Since 2.9.0 rc1:
463

    
464
- Fix bug in cfgupgrade that led to failure when upgrading from 2.8 with
465
  at least one DRBD instance.
466
- Fix bug in cfgupgrade that led to an invalid 2.8 configuration after
467
  downgrading.
468

    
469

    
470
Version 2.9.0 rc1
471
-----------------
472

    
473
*(Released Tue, 1 Oct 2013)*
474

    
475
The first release candidate in the 2.9 series. Since 2.9.0 beta1:
476

    
477
- various bug fixes
478
- update of the documentation, in particular installation instructions
479
- merging of LD_* constants into DT_* constants
480
- python style changes to be compatible with newer versions of pylint
481

    
482

    
483
Version 2.9.0 beta1
484
-------------------
485

    
486
*(Released Thu, 29 Aug 2013)*
487

    
488
This was the first beta release of the 2.9 series. All important changes
489
are listed in the latest 2.9 entry.
490

    
491

    
492
Version 2.8.4
493
-------------
494

    
495
*(Released Thu, 23 Jan 2014)*
496

    
497
- Change the list separator for the usb_devices parameter from comma to space.
498
  Commas could not work because they are already the hypervisor option
499
  separator (Issue 649)
500
- Add support for blktap2 file-driver (Issue 638)
501
- Add network tag definitions to the haskell codebase (Issue 641)
502
- Fix RAPI network tag handling
503
- Add the network tags to the tags searched by gnt-cluster search-tags
504
- Fix caching bug preventing jobs from being cancelled
505
- Start-master/stop-master was always failing if ConfD was disabled. (Issue 685)
506

    
507

    
508
Version 2.8.3
509
-------------
510

    
511
*(Released Thu, 12 Dec 2013)*
512

    
513
- Fixed Luxi daemon socket permissions after master-failover
514
- Improve IP version detection code directly checking for colons rather than
515
  passing the family from the cluster object
516
- Fix NODE/NODE_RES locking in LUInstanceCreate by not acquiring NODE_RES locks
517
  opportunistically anymore (Issue 622)
518
- Allow link local IPv6 gateways (Issue 624)
519
- Fix error printing (Issue 616)
520
- Fix a bug in InstanceSetParams concerning names: in case no name is passed in
521
  disk modifications, keep the old one. If name=none then set disk name to
522
  None.
523
- Update build_chroot script to work with the latest hackage packages
524
- Add a packet number limit to "fping" in master-ip-setup (Issue 630)
525
- Fix evacuation out of drained node (Issue 615)
526
- Add default file_driver if missing (Issue 571)
527
- Fix job error message after unclean master shutdown (Issue 618)
528
- Lock group(s) when creating instances (Issue 621)
529
- SetDiskID() before accepting an instance (Issue 633)
530
- Allow the ext template disks to receive arbitrary parameters, both at creation
531
  time and while being modified
532
- Xen handle domain shutdown (future proofing cherry-pick)
533
- Refactor reading live data in htools (future proofing cherry-pick)
534

    
535

    
536
Version 2.8.2
537
-------------
538

    
539
*(Released Thu, 07 Nov 2013)*
540

    
541
- DRBD: ensure peers are UpToDate for dual-primary
542
- Improve error message for replace-disks
543
- More dependency checks at configure time
544
- Placate warnings on ganeti.outils_unittest.py
545

    
546

    
547
Version 2.8.1
548
-------------
549

    
550
*(Released Thu, 17 Oct 2013)*
551

    
552
- Correctly start/stop luxid during gnt-cluster master-failover
553
- Don't attempt IPv6 ssh in case of IPv4 cluster (Issue 595)
554
- Fix path for the job queue serial file
555
- Improved harep man page
556
- Minor documentation improvements
557

    
558

    
559
Version 2.8.0
560
-------------
561

    
562
*(Released Mon, 30 Sep 2013)*
563

    
564
Incompatible/important changes
565
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
566

    
567
- Instance policy can contain multiple instance specs, as described in
568
  the “Constrained instance sizes” section of :doc:`Partitioned Ganeti
569
  <design-partitioned>`. As a consequence, it's not possible to partially change
570
  or override instance specs. Bounding specs (min and max) can be specified as a
571
  whole using the new option ``--ipolicy-bounds-specs``, while standard
572
  specs use the new option ``--ipolicy-std-specs``.
573
- The output of the info command of gnt-cluster, gnt-group, gnt-node,
574
  gnt-instance is a valid YAML object.
575
- hail now honors network restrictions when allocating nodes. This led to an
576
  update of the IAllocator protocol. See the IAllocator documentation for
577
  details.
578
- confd now only answers static configuration request over the network. luxid
579
  was extracted, listens on the local LUXI socket and responds to live queries.
580
  This allows finer grained permissions if using separate users.
581

    
582
New features
583
~~~~~~~~~~~~
584

    
585
- The :doc:`Remote API <rapi>` daemon now supports a command line flag
586
  to always require authentication, ``--require-authentication``. It can
587
  be specified in ``$sysconfdir/default/ganeti``.
588
- A new cluster attribute 'enabled_disk_templates' is introduced. It will
589
  be used to manage the disk templates to be used by instances in the cluster.
590
  Initially, it will be set to a list that includes plain, drbd, if they were
591
  enabled by specifying a volume group name, and file and sharedfile, if those
592
  were enabled at configure time. Additionally, it will include all disk
593
  templates that are currently used by instances. The order of disk templates
594
  will be based on Ganeti's history of supporting them. In the future, the
595
  first entry of the list will be used as a default disk template on instance
596
  creation.
597
- ``cfgupgrade`` now supports a ``--downgrade`` option to bring the
598
  configuration back to the previous stable version.
599
- Disk templates in group ipolicy can be restored to the default value.
600
- Initial support for diskless instances and virtual clusters in QA.
601
- More QA and unit tests for instance policies.
602
- Every opcode now contains a reason trail (visible through ``gnt-job info``)
603
  describing why the opcode itself was executed.
604
- The monitoring daemon is now available. It allows users to query the cluster
605
  for obtaining information about the status of the system. The daemon is only
606
  responsible for providing the information over the network: the actual data
607
  gathering is performed by data collectors (currently, only the DRBD status
608
  collector is available).
609
- In order to help developers work on Ganeti, a new script
610
  (``devel/build_chroot``) is provided, for building a chroot that contains all
611
  the required development libraries and tools for compiling Ganeti on a Debian
612
  Squeeze system.
613
- A new tool, ``harep``, for performing self-repair and recreation of instances
614
  in Ganeti has been added.
615
- Split queries are enabled for tags, network, exports, cluster info, groups,
616
  jobs, nodes.
617
- New command ``show-ispecs-cmd`` for ``gnt-cluster`` and ``gnt-group``.
618
  It prints the command line to set the current policies, to ease
619
  changing them.
620
- Add the ``vnet_hdr`` HV parameter for KVM, to control whether the tap
621
  devices for KVM virtio-net interfaces will get created with VNET_HDR
622
  (IFF_VNET_HDR) support. If set to false, it disables offloading on the
623
  virtio-net interfaces, which prevents host kernel tainting and log
624
  flooding, when dealing with broken or malicious virtio-net drivers.
625
  It's set to true by default.
626
- Instance failover now supports a ``--cleanup`` parameter for fixing previous
627
  failures.
628
- Support 'viridian' parameter in Xen HVM
629
- Support DSA SSH keys in bootstrap
630
- To simplify the work of packaging frameworks that want to add the needed users
631
  and groups in a split-user setup themselves, at build time three files in
632
  ``doc/users`` will be generated. The ``groups`` files contains, one per line,
633
  the groups to be generated, the ``users`` file contains, one per line, the
634
  users to be generated, optionally followed by their primary group, where
635
  important. The ``groupmemberships`` file contains, one per line, additional
636
  user-group membership relations that need to be established. The syntax of
637
  these files will remain stable in all future versions.
638

    
639

    
640
New dependencies
641
~~~~~~~~~~~~~~~~
642
The following new dependencies have been added:
643

    
644
For Haskell:
645
- The ``curl`` library is not optional anymore for compiling the Haskell code.
646
- ``snap-server`` library (if monitoring is enabled).
647

    
648
For Python:
649
- The minimum Python version needed to run Ganeti is now 2.6.
650
- ``yaml`` library (only for running the QA).
651

    
652
Since 2.8.0 rc3
653
~~~~~~~~~~~~~~~
654
- Perform proper cleanup on termination of Haskell daemons
655
- Fix corner-case in handling of remaining retry time
656

    
657

    
658
Version 2.8.0 rc3
659
-----------------
660

    
661
*(Released Tue, 17 Sep 2013)*
662

    
663
- To simplify the work of packaging frameworks that want to add the needed users
664
  and groups in a split-user setup themselves, at build time three files in
665
  ``doc/users`` will be generated. The ``groups`` files contains, one per line,
666
  the groups to be generated, the ``users`` file contains, one per line, the
667
  users to be generated, optionally followed by their primary group, where
668
  important. The ``groupmemberships`` file contains, one per line, additional
669
  user-group membership relations that need to be established. The syntax of
670
  these files will remain stable in all future versions.
671
- Add a default to file-driver when unspecified over RAPI (Issue 571)
672
- Mark the DSA host pubkey as optional, and remove it during config downgrade
673
  (Issue 560)
674
- Some documentation fixes
675

    
676

    
677
Version 2.8.0 rc2
678
-----------------
679

    
680
*(Released Tue, 27 Aug 2013)*
681

    
682
The second release candidate of the 2.8 series. Since 2.8.0. rc1:
683

    
684
- Support 'viridian' parameter in Xen HVM (Issue 233)
685
- Include VCS version in ``gnt-cluster version``
686
- Support DSA SSH keys in bootstrap (Issue 338)
687
- Fix batch creation of instances
688
- Use FQDN to check master node status (Issue 551)
689
- Make the DRBD collector more failure-resilient
690

    
691

    
692
Version 2.8.0 rc1
693
-----------------
694

    
695
*(Released Fri, 2 Aug 2013)*
696

    
697
The first release candidate of the 2.8 series. Since 2.8.0 beta1:
698

    
699
- Fix upgrading/downgrading from 2.7
700
- Increase maximum RAPI message size
701
- Documentation updates
702
- Split ``confd`` between ``luxid`` and ``confd``
703
- Merge 2.7 series up to the 2.7.1 release
704
- Allow the ``modify_etc_hosts`` option to be changed
705
- Add better debugging for ``luxid`` queries
706
- Expose bulk parameter for GetJobs in RAPI client
707
- Expose missing ``network`` fields in RAPI
708
- Add some ``cluster verify`` tests
709
- Some unittest fixes
710
- Fix a malfunction in ``hspace``'s tiered allocation
711
- Fix query compatibility between haskell and python implementations
712
- Add the ``vnet_hdr`` HV parameter for KVM
713
- Add ``--cleanup`` to instance failover
714
- Change the connected groups format in ``gnt-network info`` output; it
715
  was previously displayed as a raw list by mistake. (Merged from 2.7)
716

    
717

    
718
Version 2.8.0 beta1
719
-------------------
720

    
721
*(Released Mon, 24 Jun 2013)*
722

    
723
This was the first beta release of the 2.8 series. All important changes
724
are listed in the latest 2.8 entry.
725

    
726

    
727
Version 2.7.2
728
-------------
729

    
730
*(Released Thu, 26 Sep 2013)*
731

    
732
- Change the connected groups format in ``gnt-network info`` output; it
733
  was previously displayed as a raw list by mistake
734
- Check disk template in right dict when copying
735
- Support multi-instance allocs without iallocator
736
- Fix some errors in the documentation
737
- Fix formatting of tuple in an error message
738

    
739

    
740
Version 2.7.1
741
-------------
742

    
743
*(Released Thu, 25 Jul 2013)*
744

    
745
- Add logrotate functionality in daemon-util
746
- Add logrotate example file
747
- Add missing fields to network queries over rapi
748
- Fix network object timestamps
749
- Add support for querying network timestamps
750
- Fix a typo in the example crontab
751
- Fix a documentation typo
752

    
753

    
754
Version 2.7.0
755
-------------
756

    
757
*(Released Thu, 04 Jul 2013)*
758

    
759
Incompatible/important changes
760
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
761

    
762
- Instance policies for disk size were documented to be on a per-disk
763
  basis, but hail applied them to the sum of all disks. This has been
764
  fixed.
765
- ``hbal`` will now exit with status 0 if, during job execution over
766
  LUXI, early exit has been requested and all jobs are successful;
767
  before, exit status 1 was used, which cannot be differentiated from
768
  "job error" case
769
- Compatibility with newer versions of rbd has been fixed
770
- ``gnt-instance batch-create`` has been changed to use the bulk create
771
  opcode from Ganeti. This lead to incompatible changes in the format of
772
  the JSON file. It's now not a custom dict anymore but a dict
773
  compatible with the ``OpInstanceCreate`` opcode.
774
- Parent directories for file storage need to be listed in
775
  ``$sysconfdir/ganeti/file-storage-paths`` now. ``cfgupgrade`` will
776
  write the file automatically based on old configuration values, but it
777
  can not distribute it across all nodes and the file contents should be
778
  verified. Use ``gnt-cluster copyfile
779
  $sysconfdir/ganeti/file-storage-paths`` once the cluster has been
780
  upgraded. The reason for requiring this list of paths now is that
781
  before it would have been possible to inject new paths via RPC,
782
  allowing files to be created in arbitrary locations. The RPC protocol
783
  is protected using SSL/X.509 certificates, but as a design principle
784
  Ganeti does not permit arbitrary paths to be passed.
785
- The parsing of the variants file for OSes (see
786
  :manpage:`ganeti-os-interface(7)`) has been slightly changed: now empty
787
  lines and comment lines (starting with ``#``) are ignored for better
788
  readability.
789
- The ``setup-ssh`` tool added in Ganeti 2.2 has been replaced and is no
790
  longer available. ``gnt-node add`` now invokes a new tool on the
791
  destination node, named ``prepare-node-join``, to configure the SSH
792
  daemon. Paramiko is no longer necessary to configure nodes' SSH
793
  daemons via ``gnt-node add``.
794
- Draining (``gnt-cluster queue drain``) and un-draining the job queue
795
  (``gnt-cluster queue undrain``) now affects all nodes in a cluster and
796
  the flag is not reset after a master failover.
797
- Python 2.4 has *not* been tested with this release. Using 2.6 or above
798
  is recommended. 2.6 will be mandatory from the 2.8 series.
799

    
800

    
801
New features
802
~~~~~~~~~~~~
803

    
804
- New network management functionality to support automatic allocation
805
  of IP addresses and managing of network parameters. See
806
  :manpage:`gnt-network(8)` for more details.
807
- New external storage backend, to allow managing arbitrary storage
808
  systems external to the cluster. See
809
  :manpage:`ganeti-extstorage-interface(7)`.
810
- New ``exclusive-storage`` node parameter added, restricted to
811
  nodegroup level. When it's set to true, physical disks are assigned in
812
  an exclusive fashion to instances, as documented in :doc:`Partitioned
813
  Ganeti <design-partitioned>`.  Currently, only instances using the
814
  ``plain`` disk template are supported.
815
- The KVM hypervisor has been updated with many new hypervisor
816
  parameters, including a generic one for passing arbitrary command line
817
  values. See a complete list in :manpage:`gnt-instance(8)`. It is now
818
  compatible up to qemu 1.4.
819
- A new tool, called ``mon-collector``, is the stand-alone executor of
820
  the data collectors for a monitoring system. As of this version, it
821
  just includes the DRBD data collector, that can be executed by calling
822
  ``mon-collector`` using the ``drbd`` parameter. See
823
  :manpage:`mon-collector(7)`.
824
- A new user option, :pyeval:`rapi.RAPI_ACCESS_READ`, has been added
825
  for RAPI users. It allows granting permissions to query for
826
  information to a specific user without giving
827
  :pyeval:`rapi.RAPI_ACCESS_WRITE` permissions.
828
- A new tool named ``node-cleanup`` has been added. It cleans remains of
829
  a cluster from a machine by stopping all daemons, removing
830
  certificates and ssconf files. Unless the ``--no-backup`` option is
831
  given, copies of the certificates are made.
832
- Instance creations now support the use of opportunistic locking,
833
  potentially speeding up the (parallel) creation of multiple instances.
834
  This feature is currently only available via the :doc:`RAPI
835
  <rapi>` interface and when an instance allocator is used. If the
836
  ``opportunistic_locking`` parameter is set the opcode will try to
837
  acquire as many locks as possible, but will not wait for any locks
838
  held by other opcodes. If not enough resources can be found to
839
  allocate the instance, the temporary error code
840
  :pyeval:`errors.ECODE_TEMP_NORES` is returned. The operation can be
841
  retried thereafter, with or without opportunistic locking.
842
- New experimental linux-ha resource scripts.
843
- Restricted-commands support: ganeti can now be asked (via command line
844
  or rapi) to perform commands on a node. These are passed via ganeti
845
  RPC rather than ssh. This functionality is restricted to commands
846
  specified on the ``$sysconfdir/ganeti/restricted-commands`` for security
847
  reasons. The file is not copied automatically.
848

    
849

    
850
Misc changes
851
~~~~~~~~~~~~
852

    
853
- Diskless instances are now externally mirrored (Issue 237). This for
854
  now has only been tested in conjunction with explicit target nodes for
855
  migration/failover.
856
- Queries not needing locks or RPC access to the node can now be
857
  performed by the confd daemon, making them independent from jobs, and
858
  thus faster to execute. This is selectable at configure time.
859
- The functionality for allocating multiple instances at once has been
860
  overhauled and is now also available through :doc:`RAPI <rapi>`.
861

    
862
There are no significant changes from version 2.7.0~rc3.
863

    
864

    
865
Version 2.7.0 rc3
866
-----------------
867

    
868
*(Released Tue, 25 Jun 2013)*
869

    
870
- Fix permissions on the confd query socket (Issue 477)
871
- Fix permissions on the job archive dir (Issue 498)
872
- Fix handling of an internal exception in replace-disks (Issue 472)
873
- Fix gnt-node info handling of shortened names (Issue 497)
874
- Fix gnt-instance grow-disk when wiping is enabled
875
- Documentation improvements, and support for newer pandoc
876
- Fix hspace honoring ipolicy for disks (Issue 484)
877
- Improve handling of the ``kvm_extra`` HV parameter
878

    
879

    
880
Version 2.7.0 rc2
881
-----------------
882

    
883
*(Released Fri, 24 May 2013)*
884

    
885
- ``devel/upload`` now works when ``/var/run`` on the target nodes is a
886
  symlink.
887
- Disks added through ``gnt-instance modify`` or created through
888
  ``gnt-instance recreate-disks`` are wiped, if the
889
  ``prealloc_wipe_disks`` flag is set.
890
- If wiping newly created disks fails, the disks are removed. Also,
891
  partial failures in creating disks through ``gnt-instance modify``
892
  triggers a cleanup of the partially-created disks.
893
- Removing the master IP address doesn't fail if the address has been
894
  already removed.
895
- Fix ownership of the OS log dir
896
- Workaround missing SO_PEERCRED constant (Issue 191)
897

    
898

    
899
Version 2.7.0 rc1
900
-----------------
901

    
902
*(Released Fri, 3 May 2013)*
903

    
904
This was the first release candidate of the 2.7 series. Since beta3:
905

    
906
- Fix kvm compatibility with qemu 1.4 (Issue 389)
907
- Documentation updates (admin guide, upgrade notes, install
908
  instructions) (Issue 372)
909
- Fix gnt-group list nodes and instances count (Issue 436)
910
- Fix compilation without non-mandatory libraries (Issue 441)
911
- Fix xen-hvm hypervisor forcing nics to type 'ioemu' (Issue 247)
912
- Make confd logging more verbose at INFO level (Issue 435)
913
- Improve "networks" documentation in :manpage:`gnt-instance(8)`
914
- Fix failure path for instance storage type conversion (Issue 229)
915
- Update htools text backend documentation
916
- Improve the renew-crypto section of :manpage:`gnt-cluster(8)`
917
- Disable inter-cluster instance move for file-based instances, because
918
  it is dependant on instance export, which is not supported for
919
  file-based instances. (Issue 414)
920
- Fix gnt-job crashes on non-ascii characters (Issue 427)
921
- Fix volume group checks on non-vm-capable nodes (Issue 432)
922

    
923

    
924
Version 2.7.0 beta3
925
-------------------
926

    
927
*(Released Mon, 22 Apr 2013)*
928

    
929
This was the third beta release of the 2.7 series. Since beta2:
930

    
931
- Fix hail to verify disk instance policies on a per-disk basis (Issue 418).
932
- Fix data loss on wrong usage of ``gnt-instance move``
933
- Properly export errors in confd-based job queries
934
- Add ``users-setup`` tool
935
- Fix iallocator protocol to report 0 as a disk size for diskless
936
  instances. This avoids hail breaking when a diskless instance is
937
  present.
938
- Fix job queue directory permission problem that made confd job queries
939
  fail. This requires running an ``ensure-dirs --full-run`` on upgrade
940
  for access to archived jobs (Issue 406).
941
- Limit the sizes of networks supported by ``gnt-network`` to something
942
  between a ``/16`` and a ``/30`` to prevent memory bloat and crashes.
943
- Fix bugs in instance disk template conversion
944
- Fix GHC 7 compatibility
945
- Fix ``burnin`` install path (Issue 426).
946
- Allow very small disk grows (Issue 347).
947
- Fix a ``ganeti-noded`` memory bloat introduced in 2.5, by making sure
948
  that noded doesn't import masterd code (Issue 419).
949
- Make sure the default metavg at cluster init is the same as the vg, if
950
  unspecified (Issue 358).
951
- Fix cleanup of partially created disks (part of Issue 416)
952

    
953

    
954
Version 2.7.0 beta2
955
-------------------
956

    
957
*(Released Tue, 2 Apr 2013)*
958

    
959
This was the second beta release of the 2.7 series. Since beta1:
960

    
961
- Networks no longer have a "type" slot, since this information was
962
  unused in Ganeti: instead of it tags should be used.
963
- The rapi client now has a ``target_node`` option to MigrateInstance.
964
- Fix early exit return code for hbal (Issue 386).
965
- Fix ``gnt-instance migrate/failover -n`` (Issue 396).
966
- Fix ``rbd showmapped`` output parsing (Issue 312).
967
- Networks are now referenced indexed by UUID, rather than name. This
968
  will require running cfgupgrade, from 2.7.0beta1, if networks are in
969
  use.
970
- The OS environment now includes network information.
971
- Deleting of a network is now disallowed if any instance nic is using
972
  it, to prevent dangling references.
973
- External storage is now documented in man pages.
974
- The exclusive_storage flag can now only be set at nodegroup level.
975
- Hbal can now submit an explicit priority with its jobs.
976
- Many network related locking fixes.
977
- Bump up the required pylint version to 0.25.1.
978
- Fix the ``no_remember`` option in RAPI client.
979
- Many ipolicy related tests, qa, and fixes.
980
- Many documentation improvements and fixes.
981
- Fix building with ``--disable-file-storage``.
982
- Fix ``-q`` option in htools, which was broken if passed more than
983
  once.
984
- Some haskell/python interaction improvements and fixes.
985
- Fix iallocator in case of missing LVM storage.
986
- Fix confd config load in case of ``--no-lvm-storage``.
987
- The confd/query functionality is now mentioned in the security
988
  documentation.
989

    
990

    
991
Version 2.7.0 beta1
992
-------------------
993

    
994
*(Released Wed, 6 Feb 2013)*
995

    
996
This was the first beta release of the 2.7 series. All important changes
997
are listed in the latest 2.7 entry.
998

    
999

    
1000
Version 2.6.2
1001
-------------
1002

    
1003
*(Released Fri, 21 Dec 2012)*
1004

    
1005
Important behaviour change: hbal won't rebalance anymore instances which
1006
have the ``auto_balance`` attribute set to false. This was the intention
1007
all along, but until now it only skipped those from the N+1 memory
1008
reservation (DRBD-specific).
1009

    
1010
A significant number of bug fixes in this release:
1011

    
1012
- Fixed disk adoption interaction with ipolicy checks.
1013
- Fixed networking issues when instances are started, stopped or
1014
  migrated, by forcing the tap device's MAC prefix to "fe" (issue 217).
1015
- Fixed the warning in cluster verify for shared storage instances not
1016
  being redundant.
1017
- Fixed removal of storage directory on shared file storage (issue 262).
1018
- Fixed validation of LVM volume group name in OpClusterSetParams
1019
  (``gnt-cluster modify``) (issue 285).
1020
- Fixed runtime memory increases (``gnt-instance modify -m``).
1021
- Fixed live migration under Xen's ``xl`` mode.
1022
- Fixed ``gnt-instance console`` with ``xl``.
1023
- Fixed building with newer Haskell compiler/libraries.
1024
- Fixed PID file writing in Haskell daemons (confd); this prevents
1025
  restart issues if confd was launched manually (outside of
1026
  ``daemon-util``) while another copy of it was running
1027
- Fixed a type error when doing live migrations with KVM (issue 297) and
1028
  the error messages for failing migrations have been improved.
1029
- Fixed opcode validation for the out-of-band commands (``gnt-node
1030
  power``).
1031
- Fixed a type error when unsetting OS hypervisor parameters (issue
1032
  311); now it's possible to unset all OS-specific hypervisor
1033
  parameters.
1034
- Fixed the ``dry-run`` mode for many operations: verification of
1035
  results was over-zealous but didn't take into account the ``dry-run``
1036
  operation, resulting in "wrong" failures.
1037
- Fixed bash completion in ``gnt-job list`` when the job queue has
1038
  hundreds of entries; especially with older ``bash`` versions, this
1039
  results in significant CPU usage.
1040

    
1041
And lastly, a few other improvements have been made:
1042

    
1043
- Added option to force master-failover without voting (issue 282).
1044
- Clarified error message on lock conflict (issue 287).
1045
- Logging of newly submitted jobs has been improved (issue 290).
1046
- Hostname checks have been made uniform between instance rename and
1047
  create (issue 291).
1048
- The ``--submit`` option is now supported by ``gnt-debug delay``.
1049
- Shutting down the master daemon by sending SIGTERM now stops it from
1050
  processing jobs waiting for locks; instead, those jobs will be started
1051
  once again after the master daemon is started the next time (issue
1052
  296).
1053
- Support for Xen's ``xl`` program has been improved (besides the fixes
1054
  above).
1055
- Reduced logging noise in the Haskell confd daemon (only show one log
1056
  entry for each config reload, instead of two).
1057
- Several man page updates and typo fixes.
1058

    
1059

    
1060
Version 2.6.1
1061
-------------
1062

    
1063
*(Released Fri, 12 Oct 2012)*
1064

    
1065
A small bugfix release. Among the bugs fixed:
1066

    
1067
- Fixed double use of ``PRIORITY_OPT`` in ``gnt-node migrate``, that
1068
  made the command unusable.
1069
- Commands that issue many jobs don't fail anymore just because some jobs
1070
  take so long that other jobs are archived.
1071
- Failures during ``gnt-instance reinstall`` are reflected by the exit
1072
  status.
1073
- Issue 190 fixed. Check for DRBD in cluster verify is enabled only when
1074
  DRBD is enabled.
1075
- When ``always_failover`` is set, ``--allow-failover`` is not required
1076
  in migrate commands anymore.
1077
- ``bash_completion`` works even if extglob is disabled.
1078
- Fixed bug with locks that made failover for RDB-based instances fail.
1079
- Fixed bug in non-mirrored instance allocation that made Ganeti choose
1080
  a random node instead of one based on the allocator metric.
1081
- Support for newer versions of pylint and pep8.
1082
- Hail doesn't fail anymore when trying to add an instance of type
1083
  ``file``, ``sharedfile`` or ``rbd``.
1084
- Added new Makefile target to rebuild the whole distribution, so that
1085
  all files are included.
1086

    
1087

    
1088
Version 2.6.0
1089
-------------
1090

    
1091
*(Released Fri, 27 Jul 2012)*
1092

    
1093

    
1094
.. attention:: The ``LUXI`` protocol has been made more consistent
1095
   regarding its handling of command arguments. This, however, leads to
1096
   incompatibility issues with previous versions. Please ensure that you
1097
   restart Ganeti daemons soon after the upgrade, otherwise most
1098
   ``LUXI`` calls (job submission, setting/resetting the drain flag,
1099
   pausing/resuming the watcher, cancelling and archiving jobs, querying
1100
   the cluster configuration) will fail.
1101

    
1102

    
1103
New features
1104
~~~~~~~~~~~~
1105

    
1106
Instance run status
1107
+++++++++++++++++++
1108

    
1109
The current ``admin_up`` field, which used to denote whether an instance
1110
should be running or not, has been removed. Instead, ``admin_state`` is
1111
introduced, with 3 possible values -- ``up``, ``down`` and ``offline``.
1112

    
1113
The rational behind this is that an instance being “down” can have
1114
different meanings:
1115

    
1116
- it could be down during a reboot
1117
- it could be temporarily be down for a reinstall
1118
- or it could be down because it is deprecated and kept just for its
1119
  disk
1120

    
1121
The previous Boolean state was making it difficult to do capacity
1122
calculations: should Ganeti reserve memory for a down instance? Now, the
1123
tri-state field makes it clear:
1124

    
1125
- in ``up`` and ``down`` state, all resources are reserved for the
1126
  instance, and it can be at any time brought up if it is down
1127
- in ``offline`` state, only disk space is reserved for it, but not
1128
  memory or CPUs
1129

    
1130
The field can have an extra use: since the transition between ``up`` and
1131
``down`` and vice-versus is done via ``gnt-instance start/stop``, but
1132
transition between ``offline`` and ``down`` is done via ``gnt-instance
1133
modify``, it is possible to given different rights to users. For
1134
example, owners of an instance could be allowed to start/stop it, but
1135
not transition it out of the offline state.
1136

    
1137
Instance policies and specs
1138
+++++++++++++++++++++++++++
1139

    
1140
In previous Ganeti versions, an instance creation request was not
1141
limited on the minimum size and on the maximum size just by the cluster
1142
resources. As such, any policy could be implemented only in third-party
1143
clients (RAPI clients, or shell wrappers over ``gnt-*``
1144
tools). Furthermore, calculating cluster capacity via ``hspace`` again
1145
required external input with regards to instance sizes.
1146

    
1147
In order to improve these workflows and to allow for example better
1148
per-node group differentiation, we introduced instance specs, which
1149
allow declaring:
1150

    
1151
- minimum instance disk size, disk count, memory size, cpu count
1152
- maximum values for the above metrics
1153
- and “standard” values (used in ``hspace`` to calculate the standard
1154
  sized instances)
1155

    
1156
The minimum/maximum values can be also customised at node-group level,
1157
for example allowing more powerful hardware to support bigger instance
1158
memory sizes.
1159

    
1160
Beside the instance specs, there are a few other settings belonging to
1161
the instance policy framework. It is possible now to customise, per
1162
cluster and node-group:
1163

    
1164
- the list of allowed disk templates
1165
- the maximum ratio of VCPUs per PCPUs (to control CPU oversubscription)
1166
- the maximum ratio of instance to spindles (see below for more
1167
  information) for local storage
1168

    
1169
All these together should allow all tools that talk to Ganeti to know
1170
what are the ranges of allowed values for instances and the
1171
over-subscription that is allowed.
1172

    
1173
For the VCPU/PCPU ratio, we already have the VCPU configuration from the
1174
instance configuration, and the physical CPU configuration from the
1175
node. For the spindle ratios however, we didn't track before these
1176
values, so new parameters have been added:
1177

    
1178
- a new node parameter ``spindle_count``, defaults to 1, customisable at
1179
  node group or node level
1180
- at new backend parameter (for instances), ``spindle_use`` defaults to 1
1181

    
1182
Note that spindles in this context doesn't need to mean actual
1183
mechanical hard-drives; it's just a relative number for both the node
1184
I/O capacity and instance I/O consumption.
1185

    
1186
Instance migration behaviour
1187
++++++++++++++++++++++++++++
1188

    
1189
While live-migration is in general desirable over failover, it is
1190
possible that for some workloads it is actually worse, due to the
1191
variable time of the “suspend” phase during live migration.
1192

    
1193
To allow the tools to work consistently over such instances (without
1194
having to hard-code instance names), a new backend parameter
1195
``always_failover`` has been added to control the migration/failover
1196
behaviour. When set to True, all migration requests for an instance will
1197
instead fall-back to failover.
1198

    
1199
Instance memory ballooning
1200
++++++++++++++++++++++++++
1201

    
1202
Initial support for memory ballooning has been added. The memory for an
1203
instance is no longer fixed (backend parameter ``memory``), but instead
1204
can vary between minimum and maximum values (backend parameters
1205
``minmem`` and ``maxmem``). Currently we only change an instance's
1206
memory when:
1207

    
1208
- live migrating or failing over and instance and the target node
1209
  doesn't have enough memory
1210
- user requests changing the memory via ``gnt-instance modify
1211
  --runtime-memory``
1212

    
1213
Instance CPU pinning
1214
++++++++++++++++++++
1215

    
1216
In order to control the use of specific CPUs by instance, support for
1217
controlling CPU pinning has been added for the Xen, HVM and LXC
1218
hypervisors. This is controlled by a new hypervisor parameter
1219
``cpu_mask``; details about possible values for this are in the
1220
:manpage:`gnt-instance(8)`. Note that use of the most specific (precise
1221
VCPU-to-CPU mapping) form will work well only when all nodes in your
1222
cluster have the same amount of CPUs.
1223

    
1224
Disk parameters
1225
+++++++++++++++
1226

    
1227
Another area in which Ganeti was not customisable were the parameters
1228
used for storage configuration, e.g. how many stripes to use for LVM,
1229
DRBD resync configuration, etc.
1230

    
1231
To improve this area, we've added disks parameters, which are
1232
customisable at cluster and node group level, and which allow to
1233
specify various parameters for disks (DRBD has the most parameters
1234
currently), for example:
1235

    
1236
- DRBD resync algorithm and parameters (e.g. speed)
1237
- the default VG for meta-data volumes for DRBD
1238
- number of stripes for LVM (plain disk template)
1239
- the RBD pool
1240

    
1241
These parameters can be modified via ``gnt-cluster modify -D …`` and
1242
``gnt-group modify -D …``, and are used at either instance creation (in
1243
case of LVM stripes, for example) or at disk “activation” time
1244
(e.g. resync speed).
1245

    
1246
Rados block device support
1247
++++++++++++++++++++++++++
1248

    
1249
A Rados (http://ceph.com/wiki/Rbd) storage backend has been added,
1250
denoted by the ``rbd`` disk template type. This is considered
1251
experimental, feedback is welcome. For details on configuring it, see
1252
the :doc:`install` document and the :manpage:`gnt-cluster(8)` man page.
1253

    
1254
Master IP setup
1255
+++++++++++++++
1256

    
1257
The existing master IP functionality works well only in simple setups (a
1258
single network shared by all nodes); however, if nodes belong to
1259
different networks, then the ``/32`` setup and lack of routing
1260
information is not enough.
1261

    
1262
To allow the master IP to function well in more complex cases, the
1263
system was reworked as follows:
1264

    
1265
- a master IP netmask setting has been added
1266
- the master IP activation/turn-down code was moved from the node daemon
1267
  to a separate script
1268
- whether to run the Ganeti-supplied master IP script or a user-supplied
1269
  on is a ``gnt-cluster init`` setting
1270

    
1271
Details about the location of the standard and custom setup scripts are
1272
in the man page :manpage:`gnt-cluster(8)`; for information about the
1273
setup script protocol, look at the Ganeti-supplied script.
1274

    
1275
SPICE support
1276
+++++++++++++
1277

    
1278
The `SPICE <http://www.linux-kvm.org/page/SPICE>`_ support has been
1279
improved.
1280

    
1281
It is now possible to use TLS-protected connections, and when renewing
1282
or changing the cluster certificates (via ``gnt-cluster renew-crypto``,
1283
it is now possible to specify spice or spice CA certificates. Also, it
1284
is possible to configure a password for SPICE sessions via the
1285
hypervisor parameter ``spice_password_file``.
1286

    
1287
There are also new parameters to control the compression and streaming
1288
options (e.g. ``spice_image_compression``, ``spice_streaming_video``,
1289
etc.). For details, see the man page :manpage:`gnt-instance(8)` and look
1290
for the spice parameters.
1291

    
1292
Lastly, it is now possible to see the SPICE connection information via
1293
``gnt-instance console``.
1294

    
1295
OVF converter
1296
+++++++++++++
1297

    
1298
A new tool (``tools/ovfconverter``) has been added that supports
1299
conversion between Ganeti and the `Open Virtualization Format
1300
<http://en.wikipedia.org/wiki/Open_Virtualization_Format>`_ (both to and
1301
from).
1302

    
1303
This relies on the ``qemu-img`` tool to convert the disk formats, so the
1304
actual compatibility with other virtualization solutions depends on it.
1305

    
1306
Confd daemon changes
1307
++++++++++++++++++++
1308

    
1309
The configuration query daemon (``ganeti-confd``) is now optional, and
1310
has been rewritten in Haskell; whether to use the daemon at all, use the
1311
Python (default) or the Haskell version is selectable at configure time
1312
via the ``--enable-confd`` parameter, which can take one of the
1313
``haskell``, ``python`` or ``no`` values. If not used, disabling the
1314
daemon will result in a smaller footprint; for larger systems, we
1315
welcome feedback on the Haskell version which might become the default
1316
in future versions.
1317

    
1318
If you want to use ``gnt-node list-drbd`` you need to have the Haskell
1319
daemon running. The Python version doesn't implement the new call.
1320

    
1321

    
1322
User interface changes
1323
~~~~~~~~~~~~~~~~~~~~~~
1324

    
1325
We have replaced the ``--disks`` option of ``gnt-instance
1326
replace-disks`` with a more flexible ``--disk`` option, which allows
1327
adding and removing disks at arbitrary indices (Issue 188). Furthermore,
1328
disk size and mode can be changed upon recreation (via ``gnt-instance
1329
recreate-disks``, which accepts the same ``--disk`` option).
1330

    
1331
As many people are used to a ``show`` command, we have added that as an
1332
alias to ``info`` on all ``gnt-*`` commands.
1333

    
1334
The ``gnt-instance grow-disk`` command has a new mode in which it can
1335
accept the target size of the disk, instead of the delta; this can be
1336
more safe since two runs in absolute mode will be idempotent, and
1337
sometimes it's also easier to specify the desired size directly.
1338

    
1339
Also the handling of instances with regard to offline secondaries has
1340
been improved. Instance operations should not fail because one of it's
1341
secondary nodes is offline, even though it's safe to proceed.
1342

    
1343
A new command ``list-drbd`` has been added to the ``gnt-node`` script to
1344
support debugging of DRBD issues on nodes. It provides a mapping of DRBD
1345
minors to instance name.
1346

    
1347
API changes
1348
~~~~~~~~~~~
1349

    
1350
RAPI coverage has improved, with (for example) new resources for
1351
recreate-disks, node power-cycle, etc.
1352

    
1353
Compatibility
1354
~~~~~~~~~~~~~
1355

    
1356
There is partial support for ``xl`` in the Xen hypervisor; feedback is
1357
welcome.
1358

    
1359
Python 2.7 is better supported, and after Ganeti 2.6 we will investigate
1360
whether to still support Python 2.4 or move to Python 2.6 as minimum
1361
required version.
1362

    
1363
Support for Fedora has been slightly improved; the provided example
1364
init.d script should work better on it and the INSTALL file should
1365
document the needed dependencies.
1366

    
1367
Internal changes
1368
~~~~~~~~~~~~~~~~
1369

    
1370
The deprecated ``QueryLocks`` LUXI request has been removed. Use
1371
``Query(what=QR_LOCK, ...)`` instead.
1372

    
1373
The LUXI requests :pyeval:`luxi.REQ_QUERY_JOBS`,
1374
:pyeval:`luxi.REQ_QUERY_INSTANCES`, :pyeval:`luxi.REQ_QUERY_NODES`,
1375
:pyeval:`luxi.REQ_QUERY_GROUPS`, :pyeval:`luxi.REQ_QUERY_EXPORTS` and
1376
:pyeval:`luxi.REQ_QUERY_TAGS` are deprecated and will be removed in a
1377
future version. :pyeval:`luxi.REQ_QUERY` should be used instead.
1378

    
1379
RAPI client: ``CertificateError`` now derives from
1380
``GanetiApiError``. This should make it more easy to handle Ganeti
1381
errors.
1382

    
1383
Deprecation warnings due to PyCrypto/paramiko import in
1384
``tools/setup-ssh`` have been silenced, as usually they are safe; please
1385
make sure to run an up-to-date paramiko version, if you use this tool.
1386

    
1387
The QA scripts now depend on Python 2.5 or above (the main code base
1388
still works with Python 2.4).
1389

    
1390
The configuration file (``config.data``) is now written without
1391
indentation for performance reasons; if you want to edit it, it can be
1392
re-formatted via ``tools/fmtjson``.
1393

    
1394
A number of bugs has been fixed in the cluster merge tool.
1395

    
1396
``x509`` certification verification (used in import-export) has been
1397
changed to allow the same clock skew as permitted by the cluster
1398
verification. This will remove some rare but hard to diagnose errors in
1399
import-export.
1400

    
1401

    
1402
Version 2.6.0 rc4
1403
-----------------
1404

    
1405
*(Released Thu, 19 Jul 2012)*
1406

    
1407
Very few changes from rc4 to the final release, only bugfixes:
1408

    
1409
- integrated fixes from release 2.5.2 (fix general boot flag for KVM
1410
  instance, fix CDROM booting for KVM instances)
1411
- fixed node group modification of node parameters
1412
- fixed issue in LUClusterVerifyGroup with multi-group clusters
1413
- fixed generation of bash completion to ensure a stable ordering
1414
- fixed a few typos
1415

    
1416

    
1417
Version 2.6.0 rc3
1418
-----------------
1419

    
1420
*(Released Fri, 13 Jul 2012)*
1421

    
1422
Third release candidate for 2.6. The following changes were done from
1423
rc3 to rc4:
1424

    
1425
- Fixed ``UpgradeConfig`` w.r.t. to disk parameters on disk objects.
1426
- Fixed an inconsistency in the LUXI protocol with the provided
1427
  arguments (NOT backwards compatible)
1428
- Fixed a bug with node groups ipolicy where ``min`` was greater than
1429
  the cluster ``std`` value
1430
- Implemented a new ``gnt-node list-drbd`` call to list DRBD minors for
1431
  easier instance debugging on nodes (requires ``hconfd`` to work)
1432

    
1433

    
1434
Version 2.6.0 rc2
1435
-----------------
1436

    
1437
*(Released Tue, 03 Jul 2012)*
1438

    
1439
Second release candidate for 2.6. The following changes were done from
1440
rc2 to rc3:
1441

    
1442
- Fixed ``gnt-cluster verify`` regarding ``master-ip-script`` on non
1443
  master candidates
1444
- Fixed a RAPI regression on missing beparams/memory
1445
- Fixed redistribution of files on offline nodes
1446
- Added possibility to run activate-disks even though secondaries are
1447
  offline. With this change it relaxes also the strictness on some other
1448
  commands which use activate disks internally:
1449
  * ``gnt-instance start|reboot|rename|backup|export``
1450
- Made it possible to remove safely an instance if its secondaries are
1451
  offline
1452
- Made it possible to reinstall even though secondaries are offline
1453

    
1454

    
1455
Version 2.6.0 rc1
1456
-----------------
1457

    
1458
*(Released Mon, 25 Jun 2012)*
1459

    
1460
First release candidate for 2.6. The following changes were done from
1461
rc1 to rc2:
1462

    
1463
- Fixed bugs with disk parameters and ``rbd`` templates as well as
1464
  ``instance_os_add``
1465
- Made ``gnt-instance modify`` more consistent regarding new NIC/Disk
1466
  behaviour. It supports now the modify operation
1467
- ``hcheck`` implemented to analyze cluster health and possibility of
1468
  improving health by rebalance
1469
- ``hbal`` has been improved in dealing with split instances
1470

    
1471

    
1472
Version 2.6.0 beta2
1473
-------------------
1474

    
1475
*(Released Mon, 11 Jun 2012)*
1476

    
1477
Second beta release of 2.6. The following changes were done from beta2
1478
to rc1:
1479

    
1480
- Fixed ``daemon-util`` with non-root user models
1481
- Fixed creation of plain instances with ``--no-wait-for-sync``
1482
- Fix wrong iv_names when running ``cfgupgrade``
1483
- Export more information in RAPI group queries
1484
- Fixed bug when changing instance network interfaces
1485
- Extended burnin to do NIC changes
1486
- query: Added ``<``, ``>``, ``<=``, ``>=`` comparison operators
1487
- Changed default for DRBD barriers
1488
- Fixed DRBD error reporting for syncer rate
1489
- Verify the options on disk parameters
1490

    
1491
And of course various fixes to documentation and improved unittests and
1492
QA.
1493

    
1494

    
1495
Version 2.6.0 beta1
1496
-------------------
1497

    
1498
*(Released Wed, 23 May 2012)*
1499

    
1500
First beta release of 2.6. The following changes were done from beta1 to
1501
beta2:
1502

    
1503
- integrated patch for distributions without ``start-stop-daemon``
1504
- adapted example init.d script to work on Fedora
1505
- fixed log handling in Haskell daemons
1506
- adapted checks in the watcher for pycurl linked against libnss
1507
- add partial support for ``xl`` instead of ``xm`` for Xen
1508
- fixed a type issue in cluster verification
1509
- fixed ssconf handling in the Haskell code (was breaking confd in IPv6
1510
  clusters)
1511

    
1512
Plus integrated fixes from the 2.5 branch:
1513

    
1514
- fixed ``kvm-ifup`` to use ``/bin/bash``
1515
- fixed parallel build failures
1516
- KVM live migration when using a custom keymap
1517

    
1518

    
1519
Version 2.5.2
1520
-------------
1521

    
1522
*(Released Tue, 24 Jul 2012)*
1523

    
1524
A small bugfix release, with no new features:
1525

    
1526
- fixed bash-isms in kvm-ifup, for compatibility with systems which use a
1527
  different default shell (e.g. Debian, Ubuntu)
1528
- fixed KVM startup and live migration with a custom keymap (fixes Issue
1529
  243 and Debian bug #650664)
1530
- fixed compatibility with KVM versions that don't support multiple boot
1531
  devices (fixes Issue 230 and Debian bug #624256)
1532

    
1533
Additionally, a few fixes were done to the build system (fixed parallel
1534
build failures) and to the unittests (fixed race condition in test for
1535
FileID functions, and the default enable/disable mode for QA test is now
1536
customisable).
1537

    
1538

    
1539
Version 2.5.1
1540
-------------
1541

    
1542
*(Released Fri, 11 May 2012)*
1543

    
1544
A small bugfix release.
1545

    
1546
The main issues solved are on the topic of compatibility with newer LVM
1547
releases:
1548

    
1549
- fixed parsing of ``lv_attr`` field
1550
- adapted to new ``vgreduce --removemissing`` behaviour where sometimes
1551
  the ``--force`` flag is needed
1552

    
1553
Also on the topic of compatibility, ``tools/lvmstrap`` has been changed
1554
to accept kernel 3.x too (was hardcoded to 2.6.*).
1555

    
1556
A regression present in 2.5.0 that broke handling (in the gnt-* scripts)
1557
of hook results and that also made display of other errors suboptimal
1558
was fixed; the code behaves now like 2.4 and earlier.
1559

    
1560
Another change in 2.5, the cleanup of the OS scripts environment, is too
1561
aggressive: it removed even the ``PATH`` variable, which requires the OS
1562
scripts to *always* need to export it. Since this is a bit too strict,
1563
we now export a minimal PATH, the same that we export for hooks.
1564

    
1565
The fix for issue 201 (Preserve bridge MTU in KVM ifup script) was
1566
integrated into this release.
1567

    
1568
Finally, a few other miscellaneous changes were done (no new features,
1569
just small improvements):
1570

    
1571
- Fix ``gnt-group --help`` display
1572
- Fix hardcoded Xen kernel path
1573
- Fix grow-disk handling of invalid units
1574
- Update synopsis for ``gnt-cluster repair-disk-sizes``
1575
- Accept both PUT and POST in noded (makes future upgrade to 2.6 easier)
1576

    
1577

    
1578
Version 2.5.0
1579
-------------
1580

    
1581
*(Released Thu, 12 Apr 2012)*
1582

    
1583
Incompatible/important changes and bugfixes
1584
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1585

    
1586
- The default of the ``/2/instances/[instance_name]/rename`` RAPI
1587
  resource's ``ip_check`` parameter changed from ``True`` to ``False``
1588
  to match the underlying LUXI interface.
1589
- The ``/2/nodes/[node_name]/evacuate`` RAPI resource was changed to use
1590
  body parameters, see :doc:`RAPI documentation <rapi>`. The server does
1591
  not maintain backwards-compatibility as the underlying operation
1592
  changed in an incompatible way. The RAPI client can talk to old
1593
  servers, but it needs to be told so as the return value changed.
1594
- When creating file-based instances via RAPI, the ``file_driver``
1595
  parameter no longer defaults to ``loop`` and must be specified.
1596
- The deprecated ``bridge`` NIC parameter is no longer supported. Use
1597
  ``link`` instead.
1598
- Support for the undocumented and deprecated RAPI instance creation
1599
  request format version 0 has been dropped. Use version 1, supported
1600
  since Ganeti 2.1.3 and :doc:`documented <rapi>`, instead.
1601
- Pyparsing 1.4.6 or above is required, see :doc:`installation
1602
  documentation <install>`.
1603
- The "cluster-verify" hooks are now executed per group by the
1604
  ``OP_CLUSTER_VERIFY_GROUP`` opcode. This maintains the same behavior
1605
  if you just run ``gnt-cluster verify``, which generates one opcode per
1606
  group.
1607
- The environment as passed to the OS scripts is cleared, and thus no
1608
  environment variables defined in the node daemon's environment will be
1609
  inherited by the scripts.
1610
- The :doc:`iallocator <iallocator>` mode ``multi-evacuate`` has been
1611
  deprecated.
1612
- :doc:`New iallocator modes <design-multi-reloc>` have been added to
1613
  support operations involving multiple node groups.
1614
- Offline nodes are ignored when failing over an instance.
1615
- Support for KVM version 1.0, which changed the version reporting format
1616
  from 3 to 2 digits.
1617
- TCP/IP ports used by DRBD disks are returned to a pool upon instance
1618
  removal.
1619
- ``Makefile`` is now compatible with Automake 1.11.2
1620
- Includes all bugfixes made in the 2.4 series
1621

    
1622
New features
1623
~~~~~~~~~~~~
1624

    
1625
- The ganeti-htools project has been merged into the ganeti-core source
1626
  tree and will be built as part of Ganeti (see :doc:`install-quick`).
1627
- Implemented support for :doc:`shared storage <design-shared-storage>`.
1628
- Add support for disks larger than 2 TB in ``lvmstrap`` by supporting
1629
  GPT-style partition tables (requires `parted
1630
  <http://www.gnu.org/s/parted/>`_).
1631
- Added support for floppy drive and 2nd CD-ROM drive in KVM hypervisor.
1632
- Allowed adding tags on instance creation.
1633
- Export instance tags to hooks (``INSTANCE_TAGS``, see :doc:`hooks`)
1634
- Allow instances to be started in a paused state, enabling the user to
1635
  see the complete console output on boot using the console.
1636
- Added new hypervisor flag to control default reboot behaviour
1637
  (``reboot_behavior``).
1638
- Added support for KVM keymaps (hypervisor parameter ``keymap``).
1639
- Improved out-of-band management support:
1640

    
1641
  - Added ``gnt-node health`` command reporting the health status of
1642
    nodes.
1643
  - Added ``gnt-node power`` command to manage power status of nodes.
1644
  - Added command for emergency power-off (EPO), ``gnt-cluster epo``.
1645

    
1646
- Instance migration can fall back to failover if instance is not
1647
  running.
1648
- Filters can be used when listing nodes, instances, groups and locks;
1649
  see :manpage:`ganeti(7)` manpage.
1650
- Added post-execution status as variables to :doc:`hooks <hooks>`
1651
  environment.
1652
- Instance tags are exported/imported together with the instance.
1653
- When given an explicit job ID, ``gnt-job info`` will work for archived
1654
  jobs.
1655
- Jobs can define dependencies on other jobs (not yet supported via
1656
  RAPI or command line, but used by internal commands and usable via
1657
  LUXI).
1658

    
1659
  - Lock monitor (``gnt-debug locks``) shows jobs waiting for
1660
    dependencies.
1661

    
1662
- Instance failover is now available as a RAPI resource
1663
  (``/2/instances/[instance_name]/failover``).
1664
- ``gnt-instance info`` defaults to static information if primary node
1665
  is offline.
1666
- Opcodes have a new ``comment`` attribute.
1667
- Added basic SPICE support to KVM hypervisor.
1668
- ``tools/ganeti-listrunner`` allows passing of arguments to executable.
1669

    
1670
Node group improvements
1671
~~~~~~~~~~~~~~~~~~~~~~~
1672

    
1673
- ``gnt-cluster verify`` has been modified to check groups separately,
1674
  thereby improving performance.
1675
- Node group support has been added to ``gnt-cluster verify-disks``,
1676
  which now operates per node group.
1677
- Watcher has been changed to work better with node groups.
1678

    
1679
  - One process and state file per node group.
1680
  - Slow watcher in one group doesn't block other group's watcher.
1681

    
1682
- Added new command, ``gnt-group evacuate``, to move all instances in a
1683
  node group to other groups.
1684
- Added ``gnt-instance change-group`` to move an instance to another
1685
  node group.
1686
- ``gnt-cluster command`` and ``gnt-cluster copyfile`` now support
1687
  per-group operations.
1688
- Node groups can be tagged.
1689
- Some operations switch from an exclusive to a shared lock as soon as
1690
  possible.
1691
- Instance's primary and secondary nodes' groups are now available as
1692
  query fields (``pnode.group``, ``pnode.group.uuid``, ``snodes.group``
1693
  and ``snodes.group.uuid``).
1694

    
1695
Misc
1696
~~~~
1697

    
1698
- Numerous updates to documentation and manpages.
1699

    
1700
  - :doc:`RAPI <rapi>` documentation now has detailed parameter
1701
    descriptions.
1702
  - Some opcode/job results are now also documented, see :doc:`RAPI
1703
    <rapi>`.
1704

    
1705
- A lockset's internal lock is now also visible in lock monitor.
1706
- Log messages from job queue workers now contain information about the
1707
  opcode they're processing.
1708
- ``gnt-instance console`` no longer requires the instance lock.
1709
- A short delay when waiting for job changes reduces the number of LUXI
1710
  requests significantly.
1711
- DRBD metadata volumes are overwritten with zeros during disk creation.
1712
- Out-of-band commands no longer acquire the cluster lock in exclusive
1713
  mode.
1714
- ``devel/upload`` now uses correct permissions for directories.
1715

    
1716

    
1717
Version 2.5.0 rc6
1718
-----------------
1719

    
1720
*(Released Fri, 23 Mar 2012)*
1721

    
1722
This was the sixth release candidate of the 2.5 series.
1723

    
1724

    
1725
Version 2.5.0 rc5
1726
-----------------
1727

    
1728
*(Released Mon, 9 Jan 2012)*
1729

    
1730
This was the fifth release candidate of the 2.5 series.
1731

    
1732

    
1733
Version 2.5.0 rc4
1734
-----------------
1735

    
1736
*(Released Thu, 27 Oct 2011)*
1737

    
1738
This was the fourth release candidate of the 2.5 series.
1739

    
1740

    
1741
Version 2.5.0 rc3
1742
-----------------
1743

    
1744
*(Released Wed, 26 Oct 2011)*
1745

    
1746
This was the third release candidate of the 2.5 series.
1747

    
1748

    
1749
Version 2.5.0 rc2
1750
-----------------
1751

    
1752
*(Released Tue, 18 Oct 2011)*
1753

    
1754
This was the second release candidate of the 2.5 series.
1755

    
1756

    
1757
Version 2.5.0 rc1
1758
-----------------
1759

    
1760
*(Released Tue, 4 Oct 2011)*
1761

    
1762
This was the first release candidate of the 2.5 series.
1763

    
1764

    
1765
Version 2.5.0 beta3
1766
-------------------
1767

    
1768
*(Released Wed, 31 Aug 2011)*
1769

    
1770
This was the third beta release of the 2.5 series.
1771

    
1772

    
1773
Version 2.5.0 beta2
1774
-------------------
1775

    
1776
*(Released Mon, 22 Aug 2011)*
1777

    
1778
This was the second beta release of the 2.5 series.
1779

    
1780

    
1781
Version 2.5.0 beta1
1782
-------------------
1783

    
1784
*(Released Fri, 12 Aug 2011)*
1785

    
1786
This was the first beta release of the 2.5 series.
1787

    
1788

    
1789
Version 2.4.5
1790
-------------
1791

    
1792
*(Released Thu, 27 Oct 2011)*
1793

    
1794
- Fixed bug when parsing command line parameter values ending in
1795
  backslash
1796
- Fixed assertion error after unclean master shutdown
1797
- Disable HTTP client pool for RPC, significantly reducing memory usage
1798
  of master daemon
1799
- Fixed queue archive creation with wrong permissions
1800

    
1801

    
1802
Version 2.4.4
1803
-------------
1804

    
1805
*(Released Tue, 23 Aug 2011)*
1806

    
1807
Small bug-fixes:
1808

    
1809
- Fixed documentation for importing with ``--src-dir`` option
1810
- Fixed a bug in ``ensure-dirs`` with queue/archive permissions
1811
- Fixed a parsing issue with DRBD 8.3.11 in the Linux kernel
1812

    
1813

    
1814
Version 2.4.3
1815
-------------
1816

    
1817
*(Released Fri, 5 Aug 2011)*
1818

    
1819
Many bug-fixes and a few small features:
1820

    
1821
- Fixed argument order in ``ReserveLV`` and ``ReserveMAC`` which caused
1822
  issues when you tried to add an instance with two MAC addresses in one
1823
  request
1824
- KVM: fixed per-instance stored UID value
1825
- KVM: configure bridged NICs at migration start
1826
- KVM: Fix a bug where instance will not start with never KVM versions
1827
  (>= 0.14)
1828
- Added OS search path to ``gnt-cluster info``
1829
- Fixed an issue with ``file_storage_dir`` where you were forced to
1830
  provide an absolute path, but the documentation states it is a
1831
  relative path, the documentation was right
1832
- Added a new parameter to instance stop/start called ``--no-remember``
1833
  that will make the state change to not be remembered
1834
- Implemented ``no_remember`` at RAPI level
1835
- Improved the documentation
1836
- Node evacuation: don't call IAllocator if node is already empty
1837
- Fixed bug in DRBD8 replace disks on current nodes
1838
- Fixed bug in recreate-disks for DRBD instances
1839
- Moved assertion checking locks in ``gnt-instance replace-disks``
1840
  causing it to abort with not owning the right locks for some situation
1841
- Job queue: Fixed potential race condition when cancelling queued jobs
1842
- Fixed off-by-one bug in job serial generation
1843
- ``gnt-node volumes``: Fix instance names
1844
- Fixed aliases in bash completion
1845
- Fixed a bug in reopening log files after being sent a SIGHUP
1846
- Added a flag to burnin to allow specifying VCPU count
1847
- Bugfixes to non-root Ganeti configuration
1848

    
1849

    
1850
Version 2.4.2
1851
-------------
1852

    
1853
*(Released Thu, 12 May 2011)*
1854

    
1855
Many bug-fixes and a few new small features:
1856

    
1857
- Fixed a bug related to log opening failures
1858
- Fixed a bug in instance listing with orphan instances
1859
- Fixed a bug which prevented resetting the cluster-level node parameter
1860
  ``oob_program`` to the default
1861
- Many fixes related to the ``cluster-merge`` tool
1862
- Fixed a race condition in the lock monitor, which caused failures
1863
  during (at least) creation of many instances in parallel
1864
- Improved output for gnt-job info
1865
- Removed the quiet flag on some ssh calls which prevented debugging
1866
  failures
1867
- Improved the N+1 failure messages in cluster verify by actually
1868
  showing the memory values (needed and available)
1869
- Increased lock attempt timeouts so that when executing long operations
1870
  (e.g. DRBD replace-disks) other jobs do not enter 'blocking acquire'
1871
  too early and thus prevent the use of the 'fair' mechanism
1872
- Changed instance query data (``gnt-instance info``) to not acquire
1873
  locks unless needed, thus allowing its use on locked instance if only
1874
  static information is asked for
1875
- Improved behaviour with filesystems that do not support rename on an
1876
  opened file
1877
- Fixed the behaviour of ``prealloc_wipe_disks`` cluster parameter which
1878
  kept locks on all nodes during the wipe, which is unneeded
1879
- Fixed ``gnt-watcher`` handling of errors during hooks execution
1880
- Fixed bug in ``prealloc_wipe_disks`` with small disk sizes (less than
1881
  10GiB) which caused the wipe to fail right at the end in some cases
1882
- Fixed master IP activation when doing master failover with no-voting
1883
- Fixed bug in ``gnt-node add --readd`` which allowed the re-adding of
1884
  the master node itself
1885
- Fixed potential data-loss in under disk full conditions, where Ganeti
1886
  wouldn't check correctly the return code and would consider
1887
  partially-written files 'correct'
1888
- Fixed bug related to multiple VGs and DRBD disk replacing
1889
- Added new disk parameter ``metavg`` that allows placement of the meta
1890
  device for DRBD in a different volume group
1891
- Fixed error handling in the node daemon when the system libc doesn't
1892
  have major number 6 (i.e. if ``libc.so.6`` is not the actual libc)
1893
- Fixed lock release during replace-disks, which kept cluster-wide locks
1894
  when doing disk replaces with an iallocator script
1895
- Added check for missing bridges in cluster verify
1896
- Handle EPIPE errors while writing to the terminal better, so that
1897
  piping the output to e.g. ``less`` doesn't cause a backtrace
1898
- Fixed rare case where a ^C during Luxi calls could have been
1899
  interpreted as server errors, instead of simply terminating
1900
- Fixed a race condition in LUGroupAssignNodes (``gnt-group
1901
  assign-nodes``)
1902
- Added a few more parameters to the KVM hypervisor, allowing a second
1903
  CDROM, custom disk type for CDROMs and a floppy image
1904
- Removed redundant message in instance rename when the name is given
1905
  already as a FQDN
1906
- Added option to ``gnt-instance recreate-disks`` to allow creating the
1907
  disks on new nodes, allowing recreation when the original instance
1908
  nodes are completely gone
1909
- Added option when converting disk templates to DRBD to skip waiting
1910
  for the resync, in order to make the instance available sooner
1911
- Added two new variables to the OS scripts environment (containing the
1912
  instance's nodes)
1913
- Made the root_path and optional parameter for the xen-pvm hypervisor,
1914
  to allow use of ``pvgrub`` as bootloader
1915
- Changed the instance memory modifications to only check out-of-memory
1916
  conditions on memory increases, and turned the secondary node warnings
1917
  into errors (they can still be overridden via ``--force``)
1918
- Fixed the handling of a corner case when the Python installation gets
1919
  corrupted (e.g. a bad disk) while ganeti-noded is running and we try
1920
  to execute a command that doesn't exist
1921
- Fixed a bug in ``gnt-instance move`` (LUInstanceMove) when the primary
1922
  node of the instance returned failures during instance shutdown; this
1923
  adds the option ``--ignore-consistency`` to gnt-instance move
1924

    
1925
And as usual, various improvements to the error messages, documentation
1926
and man pages.
1927

    
1928

    
1929
Version 2.4.1
1930
-------------
1931

    
1932
*(Released Wed, 09 Mar 2011)*
1933

    
1934
Emergency bug-fix release. ``tools/cfgupgrade`` was broken and overwrote
1935
the RAPI users file if run twice (even with ``--dry-run``).
1936

    
1937
The release fixes that bug (nothing else changed).
1938

    
1939

    
1940
Version 2.4.0
1941
-------------
1942

    
1943
*(Released Mon, 07 Mar 2011)*
1944

    
1945
Final 2.4.0 release. Just a few small fixes:
1946

    
1947
- Fixed RAPI node evacuate
1948
- Fixed the kvm-ifup script
1949
- Fixed internal error handling for special job cases
1950
- Updated man page to specify the escaping feature for options
1951

    
1952

    
1953
Version 2.4.0 rc3
1954
-----------------
1955

    
1956
*(Released Mon, 28 Feb 2011)*
1957

    
1958
A critical fix for the ``prealloc_wipe_disks`` feature: it is possible
1959
that this feature wiped the disks of the wrong instance, leading to loss
1960
of data.
1961

    
1962
Other changes:
1963

    
1964
- Fixed title of query field containing instance name
1965
- Expanded the glossary in the documentation
1966
- Fixed one unittest (internal issue)
1967

    
1968

    
1969
Version 2.4.0 rc2
1970
-----------------
1971

    
1972
*(Released Mon, 21 Feb 2011)*
1973

    
1974
A number of bug fixes plus just a couple functionality changes.
1975

    
1976
On the user-visible side, the ``gnt-* list`` command output has changed
1977
with respect to "special" field states. The current rc1 style of display
1978
can be re-enabled by passing a new ``--verbose`` (``-v``) flag, but in
1979
the default output mode special fields states are displayed as follows:
1980

    
1981
- Offline resource: ``*``
1982
- Unavailable/not applicable: ``-``
1983
- Data missing (RPC failure): ``?``
1984
- Unknown field: ``??``
1985

    
1986
Another user-visible change is the addition of ``--force-join`` to
1987
``gnt-node add``.
1988

    
1989
As for bug fixes:
1990

    
1991
- ``tools/cluster-merge`` has seen many fixes and is now enabled again
1992
- Fixed regression in RAPI/instance reinstall where all parameters were
1993
  required (instead of optional)
1994
- Fixed ``gnt-cluster repair-disk-sizes``, was broken since Ganeti 2.2
1995
- Fixed iallocator usage (offline nodes were not considered offline)
1996
- Fixed ``gnt-node list`` with respect to non-vm_capable nodes
1997
- Fixed hypervisor and OS parameter validation with respect to
1998
  non-vm_capable nodes
1999
- Fixed ``gnt-cluster verify`` with respect to offline nodes (mostly
2000
  cosmetic)
2001
- Fixed ``tools/listrunner`` with respect to agent-based usage
2002

    
2003

    
2004
Version 2.4.0 rc1
2005
-----------------
2006

    
2007
*(Released Fri,  4 Feb 2011)*
2008

    
2009
Many changes and fixes since the beta1 release. While there were some
2010
internal changes, the code has been mostly stabilised for the RC
2011
release.
2012

    
2013
Note: the dumb allocator was removed in this release, as it was not kept
2014
up-to-date with the IAllocator protocol changes. It is recommended to
2015
use the ``hail`` command from the ganeti-htools package.
2016

    
2017
Note: the 2.4 and up versions of Ganeti are not compatible with the
2018
0.2.x branch of ganeti-htools. You need to upgrade to
2019
ganeti-htools-0.3.0 (or later).
2020

    
2021
Regressions fixed from 2.3
2022
~~~~~~~~~~~~~~~~~~~~~~~~~~
2023

    
2024
- Fixed the ``gnt-cluster verify-disks`` command
2025
- Made ``gnt-cluster verify-disks`` work in parallel (as opposed to
2026
  serially on nodes)
2027
- Fixed disk adoption breakage
2028
- Fixed wrong headers in instance listing for field aliases
2029

    
2030
Other bugs fixed
2031
~~~~~~~~~~~~~~~~
2032

    
2033
- Fixed corner case in KVM handling of NICs
2034
- Fixed many cases of wrong handling of non-vm_capable nodes
2035
- Fixed a bug where a missing instance symlink was not possible to
2036
  recreate with any ``gnt-*`` command (now ``gnt-instance
2037
  activate-disks`` does it)
2038
- Fixed the volume group name as reported by ``gnt-cluster
2039
  verify-disks``
2040
- Increased timeouts for the import-export code, hopefully leading to
2041
  fewer aborts due network or instance timeouts
2042
- Fixed bug in ``gnt-node list-storage``
2043
- Fixed bug where not all daemons were started on cluster
2044
  initialisation, but only at the first watcher run
2045
- Fixed many bugs in the OOB implementation
2046
- Fixed watcher behaviour in presence of instances with offline
2047
  secondaries
2048
- Fixed instance list output for instances running on the wrong node
2049
- a few fixes to the cluster-merge tool, but it still cannot merge
2050
  multi-node groups (currently it is not recommended to use this tool)
2051

    
2052

    
2053
Improvements
2054
~~~~~~~~~~~~
2055

    
2056
- Improved network configuration for the KVM hypervisor
2057
- Added e1000 as a supported NIC for Xen-HVM
2058
- Improved the lvmstrap tool to also be able to use partitions, as
2059
  opposed to full disks
2060
- Improved speed of disk wiping (the cluster parameter
2061
  ``prealloc_wipe_disks``, so that it has a low impact on the total time
2062
  of instance creations
2063
- Added documentation for the OS parameters
2064
- Changed ``gnt-instance deactivate-disks`` so that it can work if the
2065
  hypervisor is not responding
2066
- Added display of blacklisted and hidden OS information in
2067
  ``gnt-cluster info``
2068
- Extended ``gnt-cluster verify`` to also validate hypervisor, backend,
2069
  NIC and node parameters, which might create problems with currently
2070
  invalid (but undetected) configuration files, but prevents validation
2071
  failures when unrelated parameters are modified
2072
- Changed cluster initialisation to wait for the master daemon to become
2073
  available
2074
- Expanded the RAPI interface:
2075

    
2076
  - Added config redistribution resource
2077
  - Added activation/deactivation of instance disks
2078
  - Added export of console information
2079

    
2080
- Implemented log file reopening on SIGHUP, which allows using
2081
  logrotate(8) for the Ganeti log files
2082
- Added a basic OOB helper script as an example
2083

    
2084

    
2085
Version 2.4.0 beta1
2086
-------------------
2087

    
2088
*(Released Fri, 14 Jan 2011)*
2089

    
2090
User-visible
2091
~~~~~~~~~~~~
2092

    
2093
- Fixed timezone issues when formatting timestamps
2094
- Added support for node groups, available via ``gnt-group`` and other
2095
  commands
2096
- Added out-of-band framework and management, see :doc:`design
2097
  document <design-oob>`
2098
- Removed support for roman numbers from ``gnt-node list`` and
2099
  ``gnt-instance list``.
2100
- Allowed modification of master network interface via ``gnt-cluster
2101
  modify --master-netdev``
2102
- Accept offline secondaries while shutting down instance disks
2103
- Added ``blockdev_prefix`` parameter to Xen PVM and HVM hypervisors
2104
- Added support for multiple LVM volume groups
2105
- Avoid sorting nodes for ``gnt-node list`` if specific nodes are
2106
  requested
2107
- Added commands to list available fields:
2108

    
2109
  - ``gnt-node list-fields``
2110
  - ``gnt-group list-fields``
2111
  - ``gnt-instance list-fields``
2112

    
2113
- Updated documentation and man pages
2114

    
2115
Integration
2116
~~~~~~~~~~~
2117

    
2118
- Moved ``rapi_users`` file into separate directory, now named
2119
  ``.../ganeti/rapi/users``, ``cfgupgrade`` moves the file and creates a
2120
  symlink
2121
- Added new tool for running commands on many machines,
2122
  ``tools/ganeti-listrunner``
2123
- Implemented more verbose result in ``OpInstanceConsole`` opcode, also
2124
  improving the ``gnt-instance console`` output
2125
- Allowed customisation of disk index separator at ``configure`` time
2126
- Export node group allocation policy to :doc:`iallocator <iallocator>`
2127
- Added support for non-partitioned md disks in ``lvmstrap``
2128
- Added script to gracefully power off KVM instances
2129
- Split ``utils`` module into smaller parts
2130
- Changed query operations to return more detailed information, e.g.
2131
  whether an information is unavailable due to an offline node. To use
2132
  this new functionality, the LUXI call ``Query`` must be used. Field
2133
  information is now stored by the master daemon and can be retrieved
2134
  using ``QueryFields``. Instances, nodes and groups can also be queried
2135
  using the new opcodes ``OpQuery`` and ``OpQueryFields`` (not yet
2136
  exposed via RAPI). The following commands make use of this
2137
  infrastructure change:
2138

    
2139
  - ``gnt-group list``
2140
  - ``gnt-group list-fields``
2141
  - ``gnt-node list``
2142
  - ``gnt-node list-fields``
2143
  - ``gnt-instance list``
2144
  - ``gnt-instance list-fields``
2145
  - ``gnt-debug locks``
2146

    
2147
Remote API
2148
~~~~~~~~~~
2149

    
2150
- New RAPI resources (see :doc:`rapi`):
2151

    
2152
  - ``/2/modify``
2153
  - ``/2/groups``
2154
  - ``/2/groups/[group_name]``
2155
  - ``/2/groups/[group_name]/assign-nodes``
2156
  - ``/2/groups/[group_name]/modify``
2157
  - ``/2/groups/[group_name]/rename``
2158
  - ``/2/instances/[instance_name]/disk/[disk_index]/grow``
2159

    
2160
- RAPI changes:
2161

    
2162
  - Implemented ``no_install`` for instance creation
2163
  - Implemented OS parameters for instance reinstallation, allowing
2164
    use of special settings on reinstallation (e.g. for preserving data)
2165

    
2166
Misc
2167
~~~~
2168

    
2169
- Added IPv6 support in import/export
2170
- Pause DRBD synchronization while wiping disks on instance creation
2171
- Updated unittests and QA scripts
2172
- Improved network parameters passed to KVM
2173
- Converted man pages from docbook to reStructuredText
2174

    
2175

    
2176
Version 2.3.1
2177
-------------
2178

    
2179
*(Released Mon, 20 Dec 2010)*
2180

    
2181
Released version 2.3.1~rc1 without any changes.
2182

    
2183

    
2184
Version 2.3.1 rc1
2185
-----------------
2186

    
2187
*(Released Wed, 1 Dec 2010)*
2188

    
2189
- impexpd: Disable OpenSSL compression in socat if possible (backport
2190
  from master, commit e90739d625b, see :doc:`installation guide
2191
  <install-quick>` for details)
2192
- Changed unittest coverage report to exclude test scripts
2193
- Added script to check version format
2194

    
2195

    
2196
Version 2.3.0
2197
-------------
2198

    
2199
*(Released Wed, 1 Dec 2010)*
2200

    
2201
Released version 2.3.0~rc1 without any changes.
2202

    
2203

    
2204
Version 2.3.0 rc1
2205
-----------------
2206

    
2207
*(Released Fri, 19 Nov 2010)*
2208

    
2209
A number of bugfixes and documentation updates:
2210

    
2211
- Update ganeti-os-interface documentation
2212
- Fixed a bug related to duplicate MACs or similar items which should be
2213
  unique
2214
- Fix breakage in OS state modify
2215
- Reinstall instance: disallow offline secondaries (fixes bug related to
2216
  OS changing but reinstall failing)
2217
- plus all the other fixes between 2.2.1 and 2.2.2
2218

    
2219

    
2220
Version 2.3.0 rc0
2221
-----------------
2222

    
2223
*(Released Tue, 2 Nov 2010)*
2224

    
2225
- Fixed clearing of the default iallocator using ``gnt-cluster modify``
2226
- Fixed master failover race with watcher
2227
- Fixed a bug in ``gnt-node modify`` which could lead to an inconsistent
2228
  configuration
2229
- Accept previously stopped instance for export with instance removal
2230
- Simplify and extend the environment variables for instance OS scripts
2231
- Added new node flags, ``master_capable`` and ``vm_capable``
2232
- Added optional instance disk wiping prior during allocation. This is a
2233
  cluster-wide option and can be set/modified using
2234
  ``gnt-cluster {init,modify} --prealloc-wipe-disks``.
2235
- Added IPv6 support, see :doc:`design document <design-2.3>` and
2236
  :doc:`install-quick`
2237
- Added a new watcher option (``--ignore-pause``)
2238
- Added option to ignore offline node on instance start/stop
2239
  (``--ignore-offline``)
2240
- Allow overriding OS parameters with ``gnt-instance reinstall``
2241
- Added ability to change node's secondary IP address using ``gnt-node
2242
  modify``
2243
- Implemented privilege separation for all daemons except
2244
  ``ganeti-noded``, see ``configure`` options
2245
- Complain if an instance's disk is marked faulty in ``gnt-cluster
2246
  verify``
2247
- Implemented job priorities (see ``ganeti(7)`` manpage)
2248
- Ignore failures while shutting down instances during failover from
2249
  offline node
2250
- Exit daemon's bootstrap process only once daemon is ready
2251
- Export more information via ``LUInstanceQuery``/remote API
2252
- Improved documentation, QA and unittests
2253
- RAPI daemon now watches ``rapi_users`` all the time and doesn't need a
2254
  restart if the file was created or changed
2255
- Added LUXI protocol version sent with each request and response,
2256
  allowing detection of server/client mismatches
2257
- Moved the Python scripts among gnt-* and ganeti-* into modules
2258
- Moved all code related to setting up SSH to an external script,
2259
  ``setup-ssh``
2260
- Infrastructure changes for node group support in future versions
2261

    
2262

    
2263
Version 2.2.2
2264
-------------
2265

    
2266
*(Released Fri, 19 Nov 2010)*
2267

    
2268
A few small bugs fixed, and some improvements to the build system:
2269

    
2270
- Fix documentation regarding conversion to drbd
2271
- Fix validation of parameters in cluster modify (``gnt-cluster modify
2272
  -B``)
2273
- Fix error handling in node modify with multiple changes
2274
- Allow remote imports without checked names
2275

    
2276

    
2277
Version 2.2.1
2278
-------------
2279

    
2280
*(Released Tue, 19 Oct 2010)*
2281

    
2282
- Disable SSL session ID cache in RPC client
2283

    
2284

    
2285
Version 2.2.1 rc1
2286
-----------------
2287

    
2288
*(Released Thu, 14 Oct 2010)*
2289

    
2290
- Fix interaction between Curl/GnuTLS and the Python's HTTP server
2291
  (thanks Apollon Oikonomopoulos!), finally allowing the use of Curl
2292
  with GnuTLS
2293
- Fix problems with interaction between Curl and Python's HTTP server,
2294
  resulting in increased speed in many RPC calls
2295
- Improve our release script to prevent breakage with older aclocal and
2296
  Python 2.6
2297

    
2298

    
2299
Version 2.2.1 rc0
2300
-----------------
2301

    
2302
*(Released Thu, 7 Oct 2010)*
2303

    
2304
- Fixed issue 125, replace hardcoded "xenvg" in ``gnt-cluster`` with
2305
  value retrieved from master
2306
- Added support for blacklisted or hidden OS definitions
2307
- Added simple lock monitor (accessible via (``gnt-debug locks``)
2308
- Added support for -mem-path in KVM hypervisor abstraction layer
2309
- Allow overriding instance parameters in tool for inter-cluster
2310
  instance moves (``tools/move-instance``)
2311
- Improved opcode summaries (e.g. in ``gnt-job list``)
2312
- Improve consistency of OS listing by sorting it
2313
- Documentation updates
2314

    
2315

    
2316
Version 2.2.0.1
2317
---------------
2318

    
2319
*(Released Fri, 8 Oct 2010)*
2320

    
2321
- Rebuild with a newer autotools version, to fix python 2.6 compatibility
2322

    
2323

    
2324
Version 2.2.0
2325
-------------
2326

    
2327
*(Released Mon, 4 Oct 2010)*
2328

    
2329
- Fixed regression in ``gnt-instance rename``
2330

    
2331

    
2332
Version 2.2.0 rc2
2333
-----------------
2334

    
2335
*(Released Wed, 22 Sep 2010)*
2336

    
2337
- Fixed OS_VARIANT variable for OS scripts
2338
- Fixed cluster tag operations via RAPI
2339
- Made ``setup-ssh`` exit with non-zero code if an error occurred
2340
- Disabled RAPI CA checks in watcher
2341

    
2342

    
2343
Version 2.2.0 rc1
2344
-----------------
2345

    
2346
*(Released Mon, 23 Aug 2010)*
2347

    
2348
- Support DRBD versions of the format "a.b.c.d"
2349
- Updated manpages
2350
- Re-introduce support for usage from multiple threads in RAPI client
2351
- Instance renames and modify via RAPI
2352
- Work around race condition between processing and archival in job
2353
  queue
2354
- Mark opcodes following failed one as failed, too
2355
- Job field ``lock_status`` was removed due to difficulties making it
2356
  work with the changed job queue in Ganeti 2.2; a better way to monitor
2357
  locks is expected for a later 2.2.x release
2358
- Fixed dry-run behaviour with many commands
2359
- Support ``ssh-agent`` again when adding nodes
2360
- Many additional bugfixes
2361

    
2362

    
2363
Version 2.2.0 rc0
2364
-----------------
2365

    
2366
*(Released Fri, 30 Jul 2010)*
2367

    
2368
Important change: the internal RPC mechanism between Ganeti nodes has
2369
changed from using a home-grown http library (based on the Python base
2370
libraries) to use the PycURL library. This requires that PycURL is
2371
installed on nodes. Please note that on Debian/Ubuntu, PycURL is linked
2372
against GnuTLS by default. cURL's support for GnuTLS had known issues
2373
before cURL 7.21.0 and we recommend using the latest cURL release or
2374
linking against OpenSSL. Most other distributions already link PycURL
2375
and cURL against OpenSSL. The command::
2376

    
2377
  python -c 'import pycurl; print pycurl.version'
2378

    
2379
can be used to determine the libraries PycURL and cURL are linked
2380
against.
2381

    
2382
Other significant changes:
2383

    
2384
- Rewrote much of the internals of the job queue, in order to achieve
2385
  better parallelism; this decouples job query operations from the job
2386
  processing, and it should allow much nicer behaviour of the master
2387
  daemon under load, and it also has uncovered some long-standing bugs
2388
  related to the job serialisation (now fixed)
2389
- Added a default iallocator setting to the cluster parameters,
2390
  eliminating the need to always pass nodes or an iallocator for
2391
  operations that require selection of new node(s)
2392
- Added experimental support for the LXC virtualization method
2393
- Added support for OS parameters, which allows the installation of
2394
  instances to pass parameter to OS scripts in order to customise the
2395
  instance
2396
- Added a hypervisor parameter controlling the migration type (live or
2397
  non-live), since hypervisors have various levels of reliability; this
2398
  has renamed the 'live' parameter to 'mode'
2399
- Added a cluster parameter ``reserved_lvs`` that denotes reserved
2400
  logical volumes, meaning that cluster verify will ignore them and not
2401
  flag their presence as errors
2402
- The watcher will now reset the error count for failed instances after
2403
  8 hours, thus allowing self-healing if the problem that caused the
2404
  instances to be down/fail to start has cleared in the meantime
2405
- Added a cluster parameter ``drbd_usermode_helper`` that makes Ganeti
2406
  check for, and warn, if the drbd module parameter ``usermode_helper``
2407
  is not consistent with the cluster-wide setting; this is needed to
2408
  make diagnose easier of failed drbd creations
2409
- Started adding base IPv6 support, but this is not yet
2410
  enabled/available for use
2411
- Rename operations (cluster, instance) will now return the new name,
2412
  which is especially useful if a short name was passed in
2413
- Added support for instance migration in RAPI
2414
- Added a tool to pre-configure nodes for the SSH setup, before joining
2415
  them to the cluster; this will allow in the future a simplified model
2416
  for node joining (but not yet fully enabled in 2.2); this needs the
2417
  paramiko python library
2418
- Fixed handling of name-resolving errors
2419
- Fixed consistency of job results on the error path
2420
- Fixed master-failover race condition when executed multiple times in
2421
  sequence
2422
- Fixed many bugs related to the job queue (mostly introduced during the
2423
  2.2 development cycle, so not all are impacting 2.1)
2424
- Fixed instance migration with missing disk symlinks
2425
- Fixed handling of unknown jobs in ``gnt-job archive``
2426
- And many other small fixes/improvements
2427

    
2428
Internal changes:
2429

    
2430
- Enhanced both the unittest and the QA coverage
2431
- Switched the opcode validation to a generic model, and extended the
2432
  validation to all opcode parameters
2433
- Changed more parts of the code that write shell scripts to use the
2434
  same class for this
2435
- Switched the master daemon to use the asyncore library for the Luxi
2436
  server endpoint
2437

    
2438

    
2439
Version 2.2.0 beta0
2440
-------------------
2441

    
2442
*(Released Thu, 17 Jun 2010)*
2443

    
2444
- Added tool (``move-instance``) and infrastructure to move instances
2445
  between separate clusters (see :doc:`separate documentation
2446
  <move-instance>` and :doc:`design document <design-2.2>`)
2447
- Added per-request RPC timeout
2448
- RAPI now requires a Content-Type header for requests with a body (e.g.
2449
  ``PUT`` or ``POST``) which must be set to ``application/json`` (see
2450
  :rfc:`2616` (HTTP/1.1), section 7.2.1)
2451
- ``ganeti-watcher`` attempts to restart ``ganeti-rapi`` if RAPI is not
2452
  reachable
2453
- Implemented initial support for running Ganeti daemons as separate
2454
  users, see configure-time flags ``--with-user-prefix`` and
2455
  ``--with-group-prefix`` (only ``ganeti-rapi`` is supported at this
2456
  time)
2457
- Instances can be removed after export (``gnt-backup export
2458
  --remove-instance``)
2459
- Self-signed certificates generated by Ganeti now use a 2048 bit RSA
2460
  key (instead of 1024 bit)
2461
- Added new cluster configuration file for cluster domain secret
2462
- Import/export now use SSL instead of SSH
2463
- Added support for showing estimated time when exporting an instance,
2464
  see the ``ganeti-os-interface(7)`` manpage and look for
2465
  ``EXP_SIZE_FD``
2466

    
2467

    
2468
Version 2.1.8
2469
-------------
2470

    
2471
*(Released Tue, 16 Nov 2010)*
2472

    
2473
Some more bugfixes. Unless critical bugs occur, this will be the last
2474
2.1 release:
2475

    
2476
- Fix case of MAC special-values
2477
- Fix mac checker regex
2478
- backend: Fix typo causing "out of range" error
2479
- Add missing --units in gnt-instance list man page
2480

    
2481

    
2482
Version 2.1.7
2483
-------------
2484

    
2485
*(Released Tue, 24 Aug 2010)*
2486

    
2487
Bugfixes only:
2488
  - Don't ignore secondary node silently on non-mirrored disk templates
2489
    (issue 113)
2490
  - Fix --master-netdev arg name in gnt-cluster(8) (issue 114)
2491
  - Fix usb_mouse parameter breaking with vnc_console (issue 109)
2492
  - Properly document the usb_mouse parameter
2493
  - Fix path in ganeti-rapi(8) (issue 116)
2494
  - Adjust error message when the ganeti user's .ssh directory is
2495
    missing
2496
  - Add same-node-check when changing the disk template to drbd
2497

    
2498

    
2499
Version 2.1.6
2500
-------------
2501

    
2502
*(Released Fri, 16 Jul 2010)*
2503

    
2504
Bugfixes only:
2505
  - Add an option to only select some reboot types during qa/burnin.
2506
    (on some hypervisors consequent reboots are not supported)
2507
  - Fix infrequent race condition in master failover. Sometimes the old
2508
    master ip address would be still detected as up for a short time
2509
    after it was removed, causing failover to fail.
2510
  - Decrease mlockall warnings when the ctypes module is missing. On
2511
    Python 2.4 we support running even if no ctypes module is installed,
2512
    but we were too verbose about this issue.
2513
  - Fix building on old distributions, on which man doesn't have a
2514
    --warnings option.
2515
  - Fix RAPI not to ignore the MAC address on instance creation
2516
  - Implement the old instance creation format in the RAPI client.
2517

    
2518

    
2519
Version 2.1.5
2520
-------------
2521

    
2522
*(Released Thu, 01 Jul 2010)*
2523

    
2524
A small bugfix release:
2525
  - Fix disk adoption: broken by strict --disk option checking in 2.1.4
2526
  - Fix batch-create: broken in the whole 2.1 series due to a lookup on
2527
    a non-existing option
2528
  - Fix instance create: the --force-variant option was ignored
2529
  - Improve pylint 0.21 compatibility and warnings with Python 2.6
2530
  - Fix modify node storage with non-FQDN arguments
2531
  - Fix RAPI client to authenticate under Python 2.6 when used
2532
    for more than 5 requests needing authentication
2533
  - Fix gnt-instance modify -t (storage) giving a wrong error message
2534
    when converting a non-shutdown drbd instance to plain
2535

    
2536

    
2537
Version 2.1.4
2538
-------------
2539

    
2540
*(Released Fri, 18 Jun 2010)*
2541

    
2542
A small bugfix release:
2543

    
2544
  - Fix live migration of KVM instances started with older Ganeti
2545
    versions which had fewer hypervisor parameters
2546
  - Fix gnt-instance grow-disk on down instances
2547
  - Fix an error-reporting bug during instance migration
2548
  - Better checking of the ``--net`` and ``--disk`` values, to avoid
2549
    silently ignoring broken ones
2550
  - Fix an RPC error reporting bug affecting, for example, RAPI client
2551
    users
2552
  - Fix bug triggered by different API version os-es on different nodes
2553
  - Fix a bug in instance startup with custom hvparams: OS level
2554
    parameters would fail to be applied.
2555
  - Fix the RAPI client under Python 2.6 (but more work is needed to
2556
    make it work completely well with OpenSSL)
2557
  - Fix handling of errors when resolving names from DNS
2558

    
2559

    
2560
Version 2.1.3
2561
-------------
2562

    
2563
*(Released Thu, 3 Jun 2010)*
2564

    
2565
A medium sized development cycle. Some new features, and some
2566
fixes/small improvements/cleanups.
2567

    
2568
Significant features
2569
~~~~~~~~~~~~~~~~~~~~
2570

    
2571
The node deamon now tries to mlock itself into memory, unless the
2572
``--no-mlock`` flag is passed. It also doesn't fail if it can't write
2573
its logs, and falls back to console logging. This allows emergency
2574
features such as ``gnt-node powercycle`` to work even in the event of a
2575
broken node disk (tested offlining the disk hosting the node's
2576
filesystem and dropping its memory caches; don't try this at home)
2577

    
2578
KVM: add vhost-net acceleration support. It can be tested with a new
2579
enough version of the kernel and of qemu-kvm.
2580

    
2581
KVM: Add instance chrooting feature. If you use privilege dropping for
2582
your VMs you can also now force them to chroot to an empty directory,
2583
before starting the emulated guest.
2584

    
2585
KVM: Add maximum migration bandwith and maximum downtime tweaking
2586
support (requires a new-enough version of qemu-kvm).
2587

    
2588
Cluster verify will now warn if the master node doesn't have the master
2589
ip configured on it.
2590

    
2591
Add a new (incompatible) instance creation request format to RAPI which
2592
supports all parameters (previously only a subset was supported, and it
2593
wasn't possible to extend the old format to accomodate all the new
2594
features. The old format is still supported, and a client can check for
2595
this feature, before using it, by checking for its presence in the
2596
``features`` RAPI resource.
2597

    
2598
Now with ancient latin support. Try it passing the ``--roman`` option to
2599
``gnt-instance info``, ``gnt-cluster info`` or ``gnt-node list``
2600
(requires the python-roman module to be installed, in order to work).
2601

    
2602
Other changes
2603
~~~~~~~~~~~~~
2604

    
2605
As usual many internal code refactorings, documentation updates, and
2606
such. Among others:
2607

    
2608
  - Lots of improvements and cleanups to the experimental Remote API
2609
    (RAPI) client library.
2610
  - A new unit test suite for the core daemon libraries.
2611
  - A fix to creating missing directories makes sure the umask is not
2612
    applied anymore. This enforces the same directory permissions
2613
    everywhere.
2614
  - Better handling terminating daemons with ctrl+c (used when running
2615
    them in debugging mode).
2616
  - Fix a race condition in live migrating a KVM instance, when stat()
2617
    on the old proc status file returned EINVAL, which is an unexpected
2618
    value.
2619
  - Fixed manpage checking with newer man and utf-8 charachters. But now
2620
    you need the en_US.UTF-8 locale enabled to build Ganeti from git.
2621

    
2622

    
2623
Version 2.1.2.1
2624
---------------
2625

    
2626
*(Released Fri, 7 May 2010)*
2627

    
2628
Fix a bug which prevented untagged KVM instances from starting.
2629

    
2630

    
2631
Version 2.1.2
2632
-------------
2633

    
2634
*(Released Fri, 7 May 2010)*
2635

    
2636
Another release with a long development cycle, during which many
2637
different features were added.
2638

    
2639
Significant features
2640
~~~~~~~~~~~~~~~~~~~~
2641

    
2642
The KVM hypervisor now can run the individual instances as non-root, to
2643
reduce the impact of a VM being hijacked due to bugs in the
2644
hypervisor. It is possible to run all instances as a single (non-root)
2645
user, to manually specify a user for each instance, or to dynamically
2646
allocate a user out of a cluster-wide pool to each instance, with the
2647
guarantee that no two instances will run under the same user ID on any
2648
given node.
2649

    
2650
An experimental RAPI client library, that can be used standalone
2651
(without the other Ganeti libraries), is provided in the source tree as
2652
``lib/rapi/client.py``. Note this client might change its interface in
2653
the future, as we iterate on its capabilities.
2654

    
2655
A new command, ``gnt-cluster renew-crypto`` has been added to easily
2656
replace the cluster's certificates and crypto keys. This might help in
2657
case they have been compromised, or have simply expired.
2658

    
2659
A new disk option for instance creation has been added that allows one
2660
to "adopt" currently existing logical volumes, with data
2661
preservation. This should allow easier migration to Ganeti from
2662
unmanaged (or managed via other software) instances.
2663

    
2664
Another disk improvement is the possibility to convert between redundant
2665
(DRBD) and plain (LVM) disk configuration for an instance. This should
2666
allow better scalability (starting with one node and growing the
2667
cluster, or shrinking a two-node cluster to one node).
2668

    
2669
A new feature that could help with automated node failovers has been
2670
implemented: if a node sees itself as offline (by querying the master
2671
candidates), it will try to shutdown (hard) all instances and any active
2672
DRBD devices. This reduces the risk of duplicate instances if an
2673
external script automatically failovers the instances on such nodes. To
2674
enable this, the cluster parameter ``maintain_node_health`` should be
2675
enabled; in the future this option (per the name) will enable other
2676
automatic maintenance features.
2677

    
2678
Instance export/import now will reuse the original instance
2679
specifications for all parameters; that means exporting an instance,
2680
deleting it and the importing it back should give an almost identical
2681
instance. Note that the default import behaviour has changed from
2682
before, where it created only one NIC; now it recreates the original
2683
number of NICs.
2684

    
2685
Cluster verify has added a few new checks: SSL certificates validity,
2686
/etc/hosts consistency across the cluster, etc.
2687

    
2688
Other changes
2689
~~~~~~~~~~~~~
2690

    
2691
As usual, many internal changes were done, documentation fixes,
2692
etc. Among others:
2693

    
2694
- Fixed cluster initialization with disabled cluster storage (regression
2695
  introduced in 2.1.1)
2696
- File-based storage supports growing the disks
2697
- Fixed behaviour of node role changes
2698
- Fixed cluster verify for some corner cases, plus a general rewrite of
2699
  cluster verify to allow future extension with more checks
2700
- Fixed log spamming by watcher and node daemon (regression introduced
2701
  in 2.1.1)
2702
- Fixed possible validation issues when changing the list of enabled
2703
  hypervisors
2704
- Fixed cleanup of /etc/hosts during node removal
2705
- Fixed RAPI response for invalid methods
2706
- Fixed bug with hashed passwords in ``ganeti-rapi`` daemon
2707
- Multiple small improvements to the KVM hypervisor (VNC usage, booting
2708
  from ide disks, etc.)
2709
- Allow OS changes without re-installation (to record a changed OS
2710
  outside of Ganeti, or to allow OS renames)
2711
- Allow instance creation without OS installation (useful for example if
2712
  the OS will be installed manually, or restored from a backup not in
2713
  Ganeti format)
2714
- Implemented option to make cluster ``copyfile`` use the replication
2715
  network
2716
- Added list of enabled hypervisors to ssconf (possibly useful for
2717
  external scripts)
2718
- Added a new tool (``tools/cfgupgrade12``) that allows upgrading from
2719
  1.2 clusters
2720
- A partial form of node re-IP is possible via node readd, which now
2721
  allows changed node primary IP
2722
- Command line utilities now show an informational message if the job is
2723
  waiting for a lock
2724
- The logs of the master daemon now show the PID/UID/GID of the
2725
  connected client
2726

    
2727

    
2728
Version 2.1.1
2729
-------------
2730

    
2731
*(Released Fri, 12 Mar 2010)*
2732

    
2733
During the 2.1.0 long release candidate cycle, a lot of improvements and
2734
changes have accumulated with were released later as 2.1.1.
2735

    
2736
Major changes
2737
~~~~~~~~~~~~~
2738

    
2739
The node evacuate command (``gnt-node evacuate``) was significantly
2740
rewritten, and as such the IAllocator protocol was changed - a new
2741
request type has been added. This unfortunate change during a stable
2742
series is designed to improve performance of node evacuations; on
2743
clusters with more than about five nodes and which are well-balanced,
2744
evacuation should proceed in parallel for all instances of the node
2745
being evacuated. As such, any existing IAllocator scripts need to be
2746
updated, otherwise the above command will fail due to the unknown
2747
request. The provided "dumb" allocator has not been updated; but the
2748
ganeti-htools package supports the new protocol since version 0.2.4.
2749

    
2750
Another important change is increased validation of node and instance
2751
names. This might create problems in special cases, if invalid host
2752
names are being used.
2753

    
2754
Also, a new layer of hypervisor parameters has been added, that sits at
2755
OS level between the cluster defaults and the instance ones. This allows
2756
customisation of virtualization parameters depending on the installed
2757
OS. For example instances with OS 'X' may have a different KVM kernel
2758
(or any other parameter) than the cluster defaults. This is intended to
2759
help managing a multiple OSes on the same cluster, without manual
2760
modification of each instance's parameters.
2761

    
2762
A tool for merging clusters, ``cluster-merge``, has been added in the
2763
tools sub-directory.
2764

    
2765
Bug fixes
2766
~~~~~~~~~
2767

    
2768
- Improved the int/float conversions that should make the code more
2769
  robust in face of errors from the node daemons
2770
- Fixed the remove node code in case of internal configuration errors
2771
- Fixed the node daemon behaviour in face of inconsistent queue
2772
  directory (e.g. read-only file-system where we can't open the files
2773
  read-write, etc.)
2774
- Fixed the behaviour of gnt-node modify for master candidate demotion;
2775
  now it either aborts cleanly or, if given the new "auto_promote"
2776
  parameter, will automatically promote other nodes as needed
2777
- Fixed compatibility with (unreleased yet) Python 2.6.5 that would
2778
  completely prevent Ganeti from working
2779
- Fixed bug for instance export when not all disks were successfully
2780
  exported
2781
- Fixed behaviour of node add when the new node is slow in starting up
2782
  the node daemon
2783
- Fixed handling of signals in the LUXI client, which should improve
2784
  behaviour of command-line scripts
2785
- Added checks for invalid node/instance names in the configuration (now
2786
  flagged during cluster verify)
2787
- Fixed watcher behaviour for disk activation errors
2788
- Fixed two potentially endless loops in http library, which led to the
2789
  RAPI daemon hanging and consuming 100% CPU in some cases
2790
- Fixed bug in RAPI daemon related to hashed passwords
2791
- Fixed bug for unintended qemu-level bridging of multi-NIC KVM
2792
  instances
2793
- Enhanced compatibility with non-Debian OSes, but not using absolute
2794
  path in some commands and allowing customisation of the ssh
2795
  configuration directory
2796
- Fixed possible future issue with new Python versions by abiding to the
2797
  proper use of ``__slots__`` attribute on classes
2798
- Added checks that should prevent directory traversal attacks
2799
- Many documentation fixes based on feedback from users
2800

    
2801
New features
2802
~~~~~~~~~~~~
2803

    
2804
- Added an "early_release" more for instance replace disks and node
2805
  evacuate, where we release locks earlier and thus allow higher
2806
  parallelism within the cluster
2807
- Added watcher hooks, intended to allow the watcher to restart other
2808
  daemons (e.g. from the ganeti-nbma project), but they can be used of
2809
  course for any other purpose
2810
- Added a compile-time disable for DRBD barriers, to increase
2811
  performance if the administrator trusts the power supply or the
2812
  storage system to not lose writes
2813
- Added the option of using syslog for logging instead of, or in
2814
  addition to, Ganeti's own log files
2815
- Removed boot restriction for paravirtual NICs for KVM, recent versions
2816
  can indeed boot from a paravirtual NIC
2817
- Added a generic debug level for many operations; while this is not
2818
  used widely yet, it allows one to pass the debug value all the way to
2819
  the OS scripts
2820
- Enhanced the hooks environment for instance moves (failovers,
2821
  migrations) where the primary/secondary nodes changed during the
2822
  operation, by adding {NEW,OLD}_{PRIMARY,SECONDARY} vars
2823
- Enhanced data validations for many user-supplied values; one important
2824
  item is the restrictions imposed on instance and node names, which
2825
  might reject some (invalid) host names
2826
- Add a configure-time option to disable file-based storage, if it's not
2827
  needed; this allows greater security separation between the master
2828
  node and the other nodes from the point of view of the inter-node RPC
2829
  protocol
2830
- Added user notification in interactive tools if job is waiting in the
2831
  job queue or trying to acquire locks
2832
- Added log messages when a job is waiting for locks
2833
- Added filtering by node tags in instance operations which admit
2834
  multiple instances (start, stop, reboot, reinstall)
2835
- Added a new tool for cluster mergers, ``cluster-merge``
2836
- Parameters from command line which are of the form ``a=b,c=d`` can now
2837
  use backslash escapes to pass in values which contain commas,
2838
  e.g. ``a=b\\c,d=e`` where the 'a' parameter would get the value
2839
  ``b,c``
2840
- For KVM, the instance name is the first parameter passed to KVM, so
2841
  that it's more visible in the process list
2842

    
2843

    
2844
Version 2.1.0
2845
-------------
2846

    
2847
*(Released Tue, 2 Mar 2010)*
2848

    
2849
Ganeti 2.1 brings many improvements with it. Major changes:
2850

    
2851
- Added infrastructure to ease automated disk repairs
2852
- Added new daemon to export configuration data in a cheaper way than
2853
  using the remote API
2854
- Instance NICs can now be routed instead of being associated with a
2855
  networking bridge
2856
- Improved job locking logic to reduce impact of jobs acquiring multiple
2857
  locks waiting for other long-running jobs
2858

    
2859
In-depth implementation details can be found in the Ganeti 2.1 design
2860
document.
2861

    
2862
Details
2863
~~~~~~~
2864

    
2865
- Added chroot hypervisor
2866
- Added more options to xen-hvm hypervisor (``kernel_path`` and
2867
  ``device_model``)
2868
- Added more options to xen-pvm hypervisor (``use_bootloader``,
2869
  ``bootloader_path`` and ``bootloader_args``)
2870
- Added the ``use_localtime`` option for the xen-hvm and kvm
2871
  hypervisors, and the default value for this has changed to false (in
2872
  2.0 xen-hvm always enabled it)
2873
- Added luxi call to submit multiple jobs in one go
2874
- Added cluster initialization option to not modify ``/etc/hosts``
2875
  file on nodes
2876
- Added network interface parameters
2877
- Added dry run mode to some LUs
2878
- Added RAPI resources:
2879

    
2880
  - ``/2/instances/[instance_name]/info``
2881
  - ``/2/instances/[instance_name]/replace-disks``
2882
  - ``/2/nodes/[node_name]/evacuate``
2883
  - ``/2/nodes/[node_name]/migrate``
2884
  - ``/2/nodes/[node_name]/role``
2885
  - ``/2/nodes/[node_name]/storage``
2886
  - ``/2/nodes/[node_name]/storage/modify``
2887
  - ``/2/nodes/[node_name]/storage/repair``
2888

    
2889
- Added OpCodes to evacuate or migrate all instances on a node
2890
- Added new command to list storage elements on nodes (``gnt-node
2891
  list-storage``) and modify them (``gnt-node modify-storage``)
2892
- Added new ssconf files with master candidate IP address
2893
  (``ssconf_master_candidates_ips``), node primary IP address
2894
  (``ssconf_node_primary_ips``) and node secondary IP address
2895
  (``ssconf_node_secondary_ips``)
2896
- Added ``ganeti-confd`` and a client library to query the Ganeti
2897
  configuration via UDP
2898
- Added ability to run hooks after cluster initialization and before
2899
  cluster destruction
2900
- Added automatic mode for disk replace (``gnt-instance replace-disks
2901
  --auto``)
2902
- Added ``gnt-instance recreate-disks`` to re-create (empty) disks
2903
  after catastrophic data-loss
2904
- Added ``gnt-node repair-storage`` command to repair damaged LVM volume
2905
  groups
2906
- Added ``gnt-instance move`` command to move instances
2907
- Added ``gnt-cluster watcher`` command to control watcher
2908
- Added ``gnt-node powercycle`` command to powercycle nodes
2909
- Added new job status field ``lock_status``
2910
- Added parseable error codes to cluster verification (``gnt-cluster
2911
  verify --error-codes``) and made output less verbose (use
2912
  ``--verbose`` to restore previous behaviour)
2913
- Added UUIDs to the main config entities (cluster, nodes, instances)
2914
- Added support for OS variants
2915
- Added support for hashed passwords in the Ganeti remote API users file
2916
  (``rapi_users``)
2917
- Added option to specify maximum timeout on instance shutdown
2918
- Added ``--no-ssh-init`` option to ``gnt-cluster init``
2919
- Added new helper script to start and stop Ganeti daemons
2920
  (``daemon-util``), with the intent to reduce the work necessary to
2921
  adjust Ganeti for non-Debian distributions and to start/stop daemons
2922
  from one place
2923
- Added more unittests
2924
- Fixed critical bug in ganeti-masterd startup
2925
- Removed the configure-time ``kvm-migration-port`` parameter, this is
2926
  now customisable at the cluster level for both the KVM and Xen
2927
  hypervisors using the new ``migration_port`` parameter
2928
- Pass ``INSTANCE_REINSTALL`` variable to OS installation script when
2929
  reinstalling an instance
2930
- Allowed ``@`` in tag names
2931
- Migrated to Sphinx (http://sphinx.pocoo.org/) for documentation
2932
- Many documentation updates
2933
- Distribute hypervisor files on ``gnt-cluster redist-conf``
2934
- ``gnt-instance reinstall`` can now reinstall multiple instances
2935
- Updated many command line parameters
2936
- Introduced new OS API version 15
2937
- No longer support a default hypervisor
2938
- Treat virtual LVs as inexistent
2939
- Improved job locking logic to reduce lock contention
2940
- Match instance and node names case insensitively
2941
- Reimplemented bash completion script to be more complete
2942
- Improved burnin
2943

    
2944

    
2945
Version 2.0.6
2946
-------------
2947

    
2948
*(Released Thu, 4 Feb 2010)*
2949

    
2950
- Fix cleaner behaviour on nodes not in a cluster (Debian bug 568105)
2951
- Fix a string formatting bug
2952
- Improve safety of the code in some error paths
2953
- Improve data validation in the master of values returned from nodes
2954

    
2955

    
2956
Version 2.0.5
2957
-------------
2958

    
2959
*(Released Thu, 17 Dec 2009)*
2960

    
2961
- Fix security issue due to missing validation of iallocator names; this
2962
  allows local and remote execution of arbitrary executables
2963
- Fix failure of gnt-node list during instance removal
2964
- Ship the RAPI documentation in the archive
2965

    
2966

    
2967
Version 2.0.4
2968
-------------
2969

    
2970
*(Released Wed, 30 Sep 2009)*
2971

    
2972
- Fixed many wrong messages
2973
- Fixed a few bugs related to the locking library
2974
- Fixed MAC checking at instance creation time
2975
- Fixed a DRBD parsing bug related to gaps in /proc/drbd
2976
- Fixed a few issues related to signal handling in both daemons and
2977
  scripts
2978
- Fixed the example startup script provided
2979
- Fixed insserv dependencies in the example startup script (patch from
2980
  Debian)
2981
- Fixed handling of drained nodes in the iallocator framework
2982
- Fixed handling of KERNEL_PATH parameter for xen-hvm (Debian bug
2983
  #528618)
2984
- Fixed error related to invalid job IDs in job polling
2985
- Fixed job/opcode persistence on unclean master shutdown
2986
- Fixed handling of partial job processing after unclean master
2987
  shutdown
2988
- Fixed error reporting from LUs, previously all errors were converted
2989
  into execution errors
2990
- Fixed error reporting from burnin
2991
- Decreased significantly the memory usage of the job queue
2992
- Optimised slightly multi-job submission
2993
- Optimised slightly opcode loading
2994
- Backported the multi-job submit framework from the development
2995
  branch; multi-instance start and stop should be faster
2996
- Added script to clean archived jobs after 21 days; this will reduce
2997
  the size of the queue directory
2998
- Added some extra checks in disk size tracking
2999
- Added an example ethers hook script
3000
- Added a cluster parameter that prevents Ganeti from modifying of
3001
  /etc/hosts
3002
- Added more node information to RAPI responses
3003
- Added a ``gnt-job watch`` command that allows following the ouput of a
3004
  job
3005
- Added a bind-address option to ganeti-rapi
3006
- Added more checks to the configuration verify
3007
- Enhanced the burnin script such that some operations can be retried
3008
  automatically
3009
- Converted instance reinstall to multi-instance model
3010

    
3011

    
3012
Version 2.0.3
3013
-------------
3014

    
3015
*(Released Fri, 7 Aug 2009)*
3016

    
3017
- Added ``--ignore-size`` to the ``gnt-instance activate-disks`` command
3018
  to allow using the pre-2.0.2 behaviour in activation, if any existing
3019
  instances have mismatched disk sizes in the configuration
3020
- Added ``gnt-cluster repair-disk-sizes`` command to check and update
3021
  any configuration mismatches for disk sizes
3022
- Added ``gnt-master cluste-failover --no-voting`` to allow master
3023
  failover to work on two-node clusters
3024
- Fixed the ``--net`` option of ``gnt-backup import``, which was
3025
  unusable
3026
- Fixed detection of OS script errors in ``gnt-backup export``
3027
- Fixed exit code of ``gnt-backup export``
3028

    
3029

    
3030
Version 2.0.2
3031
-------------
3032

    
3033
*(Released Fri, 17 Jul 2009)*
3034

    
3035
- Added experimental support for stripped logical volumes; this should
3036
  enhance performance but comes with a higher complexity in the block
3037
  device handling; stripping is only enabled when passing
3038
  ``--with-lvm-stripecount=N`` to ``configure``, but codepaths are
3039
  affected even in the non-stripped mode
3040
- Improved resiliency against transient failures at the end of DRBD
3041
  resyncs, and in general of DRBD resync checks
3042
- Fixed a couple of issues with exports and snapshot errors
3043
- Fixed a couple of issues in instance listing
3044
- Added display of the disk size in ``gnt-instance info``
3045
- Fixed checking for valid OSes in instance creation
3046
- Fixed handling of the "vcpus" parameter in instance listing and in
3047
  general of invalid parameters
3048
- Fixed http server library, and thus RAPI, to handle invalid
3049
  username/password combinations correctly; this means that now they
3050
  report unauthorized for queries too, not only for modifications,
3051
  allowing earlier detect of configuration problems
3052
- Added a new "role" node list field, equivalent to the master/master
3053
  candidate/drained/offline flags combinations
3054
- Fixed cluster modify and changes of candidate pool size
3055
- Fixed cluster verify error messages for wrong files on regular nodes
3056
- Fixed a couple of issues with node demotion from master candidate role
3057
- Fixed node readd issues
3058
- Added non-interactive mode for ``ganeti-masterd --no-voting`` startup
3059
- Added a new ``--no-voting`` option for masterfailover to fix failover
3060
  on two-nodes clusters when the former master node is unreachable
3061
- Added instance reinstall over RAPI
3062

    
3063

    
3064
Version 2.0.1
3065
-------------
3066

    
3067
*(Released Tue, 16 Jun 2009)*
3068

    
3069
- added ``-H``/``-B`` startup parameters to ``gnt-instance``, which will
3070
  allow re-adding the start in single-user option (regression from 1.2)
3071
- the watcher writes the instance status to a file, to allow monitoring
3072
  to report the instance status (from the master) based on cached
3073
  results of the watcher's queries; while this can get stale if the
3074
  watcher is being locked due to other work on the cluster, this is
3075
  still an improvement
3076
- the watcher now also restarts the node daemon and the rapi daemon if
3077
  they died
3078
- fixed the watcher to handle full and drained queue cases
3079
- hooks export more instance data in the environment, which helps if
3080
  hook scripts need to take action based on the instance's properties
3081
  (no longer need to query back into ganeti)
3082
- instance failovers when the instance is stopped do not check for free
3083
  RAM, so that failing over a stopped instance is possible in low memory
3084
  situations
3085
- rapi uses queries for tags instead of jobs (for less job traffic), and
3086
  for cluster tags it won't talk to masterd at all but read them from
3087
  ssconf
3088
- a couple of error handling fixes in RAPI
3089
- drbd handling: improved the error handling of inconsistent disks after
3090
  resync to reduce the frequency of "there are some degraded disks for
3091
  this instance" messages
3092
- fixed a bug in live migration when DRBD doesn't want to reconnect (the
3093
  error handling path called a wrong function name)
3094

    
3095

    
3096
Version 2.0.0
3097
-------------
3098

    
3099
*(Released Wed, 27 May 2009)*
3100

    
3101
- no changes from rc5
3102

    
3103

    
3104
Version 2.0 rc5
3105
---------------
3106

    
3107
*(Released Wed, 20 May 2009)*
3108

    
3109
- fix a couple of bugs (validation, argument checks)
3110
- fix ``gnt-cluster getmaster`` on non-master nodes (regression)
3111
- some small improvements to RAPI and IAllocator
3112
- make watcher automatically start the master daemon if down
3113

    
3114

    
3115
Version 2.0 rc4
3116
---------------
3117

    
3118
*(Released Mon, 27 Apr 2009)*
3119

    
3120
- change the OS list to not require locks; this helps with big clusters
3121
- fix ``gnt-cluster verify`` and ``gnt-cluster verify-disks`` when the
3122
  volume group is broken
3123
- ``gnt-instance info``, without any arguments, doesn't run for all
3124
  instances anymore; either pass ``--all`` or pass the desired
3125
  instances; this helps against mistakes on big clusters where listing
3126
  the information for all instances takes a long time
3127
- miscellaneous doc and man pages fixes
3128

    
3129

    
3130
Version 2.0 rc3
3131
---------------
3132

    
3133
*(Released Wed, 8 Apr 2009)*
3134

    
3135
- Change the internal locking model of some ``gnt-node`` commands, in
3136
  order to reduce contention (and blocking of master daemon) when
3137
  batching many creation/reinstall jobs
3138
- Fixes to Xen soft reboot
3139
- No longer build documentation at build time, instead distribute it in
3140
  the archive, in order to reduce the need for the whole docbook/rst
3141
  toolchains
3142

    
3143

    
3144
Version 2.0 rc2
3145
---------------
3146

    
3147
*(Released Fri, 27 Mar 2009)*
3148

    
3149
- Now the cfgupgrade scripts works and can upgrade 1.2.7 clusters to 2.0
3150
- Fix watcher startup sequence, improves the behaviour of busy clusters
3151
- Some other fixes in ``gnt-cluster verify``, ``gnt-instance
3152
  replace-disks``, ``gnt-instance add``, ``gnt-cluster queue``, KVM VNC
3153
  bind address and other places
3154
- Some documentation fixes and updates
3155

    
3156

    
3157
Version 2.0 rc1
3158
---------------
3159

    
3160
*(Released Mon, 2 Mar 2009)*
3161

    
3162
- More documentation updates, now all docs should be more-or-less
3163
  up-to-date
3164
- A couple of small fixes (mixed hypervisor clusters, offline nodes,
3165
  etc.)
3166
- Added a customizable HV_KERNEL_ARGS hypervisor parameter (for Xen PVM
3167
  and KVM)
3168
- Fix an issue related to $libdir/run/ganeti and cluster creation
3169

    
3170

    
3171
Version 2.0 beta2
3172
-----------------
3173

    
3174
*(Released Thu, 19 Feb 2009)*
3175

    
3176
- Xen PVM and KVM have switched the default value for the instance root
3177
  disk to the first partition on the first drive, instead of the whole
3178
  drive; this means that the OS installation scripts must be changed
3179
  accordingly
3180
- Man pages have been updated
3181
- RAPI has been switched by default to HTTPS, and the exported functions
3182
  should all work correctly
3183
- RAPI v1 has been removed
3184
- Many improvements to the KVM hypervisor
3185
- Block device errors are now better reported
3186
- Many other bugfixes and small improvements
3187

    
3188

    
3189
Version 2.0 beta1
3190
-----------------
3191

    
3192
*(Released Mon, 26 Jan 2009)*
3193

    
3194
- Version 2 is a general rewrite of the code and therefore the
3195
  differences are too many to list, see the design document for 2.0 in
3196
  the ``doc/`` subdirectory for more details
3197
- In this beta version there is not yet a migration path from 1.2 (there
3198
  will be one in the final 2.0 release)
3199
- A few significant changes are:
3200

    
3201
  - all commands are executed by a daemon (``ganeti-masterd``) and the
3202
    various ``gnt-*`` commands are just front-ends to it
3203
  - all the commands are entered into, and executed from a job queue,
3204
    see the ``gnt-job(8)`` manpage
3205
  - the RAPI daemon supports read-write operations, secured by basic
3206
    HTTP authentication on top of HTTPS
3207
  - DRBD version 0.7 support has been removed, DRBD 8 is the only
3208
    supported version (when migrating from Ganeti 1.2 to 2.0, you need
3209
    to migrate to DRBD 8 first while still running Ganeti 1.2)
3210
  - DRBD devices are using statically allocated minor numbers, which
3211
    will be assigned to existing instances during the migration process
3212
  - there is support for both Xen PVM and Xen HVM instances running on
3213
    the same cluster
3214
  - KVM virtualization is supported too
3215
  - file-based storage has been implemented, which means that it is
3216
    possible to run the cluster without LVM and DRBD storage, for
3217
    example using a shared filesystem exported from shared storage (and
3218
    still have live migration)
3219

    
3220

    
3221
Version 1.2.7
3222
-------------
3223

    
3224
*(Released Tue, 13 Jan 2009)*
3225

    
3226
- Change the default reboot type in ``gnt-instance reboot`` to "hard"
3227
- Reuse the old instance mac address by default on instance import, if
3228
  the instance name is the same.
3229
- Handle situations in which the node info rpc returns incomplete
3230
  results (issue 46)
3231
- Add checks for tcp/udp ports collisions in ``gnt-cluster verify``
3232
- Improved version of batcher:
3233

    
3234
  - state file support
3235
  - instance mac address support
3236
  - support for HVM clusters/instances
3237

    
3238
- Add an option to show the number of cpu sockets and nodes in
3239
  ``gnt-node list``
3240
- Support OSes that handle more than one version of the OS api (but do
3241
  not change the current API in any other way)
3242
- Fix ``gnt-node migrate``
3243
- ``gnt-debug`` man page
3244
- Fixes various more typos and small issues
3245
- Increase disk resync maximum speed to 60MB/s (from 30MB/s)
3246

    
3247

    
3248
Version 1.2.6
3249
-------------
3250

    
3251
*(Released Wed, 24 Sep 2008)*
3252

    
3253
- new ``--hvm-nic-type`` and ``--hvm-disk-type`` flags to control the
3254
  type of disk exported to fully virtualized instances.
3255
- provide access to the serial console of HVM instances
3256
- instance auto_balance flag, set by default. If turned off it will
3257
  avoid warnings on cluster verify if there is not enough memory to fail
3258
  over an instance. in the future it will prevent automatically failing
3259
  it over when we will support that.
3260
- batcher tool for instance creation, see ``tools/README.batcher``
3261
- ``gnt-instance reinstall --select-os`` to interactively select a new
3262
  operating system when reinstalling an instance.
3263
- when changing the memory amount on instance modify a check has been
3264
  added that the instance will be able to start. also warnings are
3265
  emitted if the instance will not be able to fail over, if auto_balance
3266
  is true.
3267
- documentation fixes
3268
- sync fields between ``gnt-instance list/modify/add/import``
3269
- fix a race condition in drbd when the sync speed was set after giving
3270
  the device a remote peer.
3271

    
3272

    
3273
Version 1.2.5
3274
-------------
3275

    
3276
*(Released Tue, 22 Jul 2008)*
3277

    
3278
- note: the allowed size and number of tags per object were reduced
3279
- fix a bug in ``gnt-cluster verify`` with inconsistent volume groups
3280
- fixed twisted 8.x compatibility
3281
- fixed ``gnt-instance replace-disks`` with iallocator
3282
- add TCP keepalives on twisted connections to detect restarted nodes
3283
- disk increase support, see ``gnt-instance grow-disk``
3284
- implement bulk node/instance query for RAPI
3285
- add tags in node/instance listing (optional)
3286
- experimental migration (and live migration) support, read the man page
3287
  for ``gnt-instance migrate``
3288
- the ``ganeti-watcher`` logs are now timestamped, and the watcher also
3289
  has some small improvements in handling its state file
3290

    
3291

    
3292
Version 1.2.4
3293
-------------
3294

    
3295
*(Released Fri, 13 Jun 2008)*
3296

    
3297
- Experimental readonly, REST-based remote API implementation;
3298
  automatically started on master node, TCP port 5080, if enabled by
3299
  ``--enable-rapi`` parameter to configure script.
3300
- Instance allocator support. Add and import instance accept a
3301
  ``--iallocator`` parameter, and call that instance allocator to decide
3302
  which node to use for the instance. The iallocator document describes
3303
  what's expected from an allocator script.
3304
- ``gnt-cluster verify`` N+1 memory redundancy checks: Unless passed the
3305
  ``--no-nplus1-mem`` option ``gnt-cluster verify`` now checks that if a
3306
  node is lost there is still enough memory to fail over the instances
3307
  that reside on it.
3308
- ``gnt-cluster verify`` hooks: it is now possible to add post-hooks to
3309
  ``gnt-cluster verify``, to check for site-specific compliance. All the
3310
  hooks will run, and their output, if any, will be displayed. Any
3311
  failing hook will make the verification return an error value.
3312
- ``gnt-cluster verify`` now checks that its peers are reachable on the
3313
  primary and secondary interfaces
3314
- ``gnt-node add`` now supports the ``--readd`` option, to readd a node
3315
  that is still declared as part of the cluster and has failed.
3316
- ``gnt-* list`` commands now accept a new ``-o +field`` way of
3317
  specifying output fields, that just adds the chosen fields to the
3318
  default ones.
3319
- ``gnt-backup`` now has a new ``remove`` command to delete an existing
3320
  export from the filesystem.
3321
- New per-instance parameters hvm_acpi, hvm_pae and hvm_cdrom_image_path
3322
  have been added. Using them you can enable/disable acpi and pae
3323
  support, and specify a path for a cd image to be exported to the
3324
  instance. These parameters as the name suggest only work on HVM
3325
  clusters.
3326
- When upgrading an HVM cluster to Ganeti 1.2.4, the values for ACPI and
3327
  PAE support will be set to the previously hardcoded values, but the
3328
  (previously hardcoded) path to the CDROM ISO image will be unset and
3329
  if required, needs to be set manually with ``gnt-instance modify``
3330
  after the upgrade.
3331
- The address to which an instance's VNC console is bound is now
3332
  selectable per-instance, rather than being cluster wide. Of course
3333
  this only applies to instances controlled via VNC, so currently just
3334
  applies to HVM clusters.
3335

    
3336

    
3337
Version 1.2.3
3338
-------------
3339

    
3340
*(Released Mon, 18 Feb 2008)*
3341

    
3342
- more tweaks to the disk activation code (especially helpful for DRBD)
3343
- change the default ``gnt-instance list`` output format, now there is
3344
  one combined status field (see the manpage for the exact values this
3345
  field will have)
3346
- some more fixes for the mac export to hooks change
3347
- make Ganeti not break with DRBD 8.2.x (which changed the version
3348
  format in ``/proc/drbd``) (issue 24)
3349
- add an upgrade tool from "remote_raid1" disk template to "drbd" disk
3350
  template, allowing migration from DRBD0.7+MD to DRBD8
3351

    
3352

    
3353
Version 1.2.2
3354
-------------
3355

    
3356
*(Released Wed, 30 Jan 2008)*
3357

    
3358
- fix ``gnt-instance modify`` breakage introduced in 1.2.1 with the HVM
3359
  support (issue 23)
3360
- add command aliases infrastructure and a few aliases
3361
- allow listing of VCPUs in the ``gnt-instance list`` and improve the
3362
  man pages and the ``--help`` option of ``gnt-node
3363
  list``/``gnt-instance list``
3364
- fix ``gnt-backup list`` with down nodes (issue 21)
3365
- change the tools location (move from $pkgdatadir to $pkglibdir/tools)
3366
- fix the dist archive and add a check for including svn/git files in
3367
  the future
3368
- some developer-related changes: improve the burnin and the QA suite,
3369
  add an upload script for testing during development
3370

    
3371

    
3372
Version 1.2.1
3373
-------------
3374

    
3375
*(Released Wed, 16 Jan 2008)*
3376

    
3377
- experimental HVM support, read the install document, section
3378
  "Initializing the cluster"
3379
- allow for the PVM hypervisor per-instance kernel and initrd paths
3380
- add a new command ``gnt-cluster verify-disks`` which uses a new
3381
  algorithm to improve the reconnection of the DRBD pairs if the device
3382
  on the secondary node has gone away
3383
- make logical volume code auto-activate LVs at disk activation time
3384
- slightly improve the speed of activating disks
3385
- allow specification of the MAC address at instance creation time, and
3386
  changing it later via ``gnt-instance modify``
3387
- fix handling of external commands that generate lots of output on
3388
  stderr
3389
- update documentation with regard to minimum version of DRBD8 supported
3390

    
3391

    
3392
Version 1.2.0
3393
-------------
3394

    
3395
*(Released Tue, 4 Dec 2007)*
3396

    
3397
- Log the ``xm create`` output to the node daemon log on failure (to
3398
  help diagnosing the error)
3399
- In debug mode, log all external commands output if failed to the logs
3400
- Change parsing of lvm commands to ignore stderr
3401

    
3402

    
3403
Version 1.2 beta3
3404
-----------------
3405

    
3406
*(Released Wed, 28 Nov 2007)*
3407

    
3408
- Another round of updates to the DRBD 8 code to deal with more failures
3409
  in the replace secondary node operation
3410
- Some more logging of failures in disk operations (lvm, drbd)
3411
- A few documentation updates
3412
- QA updates
3413

    
3414

    
3415
Version 1.2 beta2
3416
-----------------
3417

    
3418
*(Released Tue, 13 Nov 2007)*
3419

    
3420
- Change configuration file format from Python's Pickle to JSON.
3421
  Upgrading is possible using the cfgupgrade utility.
3422
- Add support for DRBD 8.0 (new disk template ``drbd``) which allows for
3423
  faster replace disks and is more stable (DRBD 8 has many improvements
3424
  compared to DRBD 0.7)
3425
- Added command line tags support (see man pages for ``gnt-instance``,
3426
  ``gnt-node``, ``gnt-cluster``)
3427
- Added instance rename support
3428
- Added multi-instance startup/shutdown
3429
- Added cluster rename support
3430
- Added ``gnt-node evacuate`` to simplify some node operations
3431
- Added instance reboot operation that can speedup reboot as compared to
3432
  stop and start
3433
- Soften the requirement that hostnames are in FQDN format
3434
- The ``ganeti-watcher`` now activates drbd pairs after secondary node
3435
  reboots
3436
- Removed dependency on debian's patched fping that uses the
3437
  non-standard ``-S`` option
3438
- Now the OS definitions are searched for in multiple, configurable
3439
  paths (easier for distros to package)
3440
- Some changes to the hooks infrastructure (especially the new
3441
  post-configuration update hook)
3442
- Other small bugfixes
3443

    
3444
.. vim: set textwidth=72 syntax=rst :
3445
.. Local Variables:
3446
.. mode: rst
3447
.. fill-column: 72
3448
.. End: