Statistics
| Branch: | Tag: | Revision:

root / NEWS @ 560ef132

History | View | Annotate | Download (126.8 kB)

1
News
2
====
3

    
4

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

    
8
*(unreleased)*
9

    
10

    
11
Version 2.11.0 alpha1
12
---------------------
13

    
14
*(unreleased)*
15

    
16
Incompatible/important changes
17
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
18

    
19
- ``gnt-node list`` no longer shows disk space information for shared file
20
  disk templates because it is not a node attribute. (For example, if you have
21
  both the file and shared file disk templates enabled, ``gnt-node list`` now
22
  only shows information about the file disk template.)
23
- The shared file disk template is now in the new 'sharedfile' storage type.
24
  As a result, ``gnt-node list-storage -t file`` now only shows information
25
  about the file disk template and you may use ``gnt-node list-storage -t
26
  sharedfile`` to query storage information for the shared file disk template.
27
- Over luxi, syntactially incorrect queries are now rejected as a whole;
28
  before, a 'SumbmitManyJobs' request was partially executed, if the outer
29
  structure of the request was syntactically correct. As the luxi protocol
30
  is internal (external applications are expected to use RAPI), the impact
31
  of this incompatible change should be limited.
32
- Queries for nodes, instances, groups, backups and networks are now
33
  exclusively done via the luxi daemon. Legacy python code was removed,
34
  as well as the --enable-split-queries configuration option.
35
- Orphan volumes errors are demoted to warnings and no longer affect the exit
36
  code of ``gnt-cluster verify``.
37
- RPC security got enhanced by using different client SSL certificates
38
  for each node. In this context 'gnt-cluster renew-crypto' got a new
39
  option '--renew-node-certificates', which renews the client
40
  certificates of all nodes. After a cluster upgrade from pre-2.11, run
41
  this to create client certificates and activate this feature.
42

    
43
New features
44
~~~~~~~~~~~~
45

    
46
- Instance moves, backups and imports can now use compression to transfer the
47
  instance data.
48
- Node groups can be configured to use an SSH port different than the
49
  default 22.
50
- Added experimental support for Gluster distributed file storage as the
51
  ``gluster`` disk template under the new ``sharedfile`` storage type through
52
  automatic management of per-node FUSE mount points. You can configure the
53
  mount point location at ``gnt-cluster init`` time by using the new
54
  ``--gluster-storage-dir`` switch.
55
- Job scheduling is now handled by luxid, and the maximal number of jobs running
56
  in parallel is a run-time parameter of the cluster.
57

    
58
New dependencies
59
~~~~~~~~~~~~~~~~
60
The following new dependencies have been added:
61

    
62
For Haskell:
63

    
64
- ``zlib`` library (http://hackage.haskell.org/package/base64-bytestring)
65

    
66
- ``base64-bytestring`` library (http://hackage.haskell.org/package/zlib),
67
  at least version 1.0.0.0
68

    
69

    
70
Version 2.10.0 rc1
71
------------------
72

    
73
*(Released Tue, 17 Dec 2013)*
74

    
75
Incompatible/important changes
76
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
77

    
78
- Adding disks with 'gnt-instance modify' now waits for the disks to sync per
79
  default. Specify --no-wait-for-sync to override this behavior.
80
- The Ganeti python code now adheres to a private-module layout. In particular,
81
  the module 'ganeti' is no longer in the python search path.
82
- On instance allocation, the iallocator now considers non-LVM storage
83
  properly. In particular, actual file storage space information is used
84
  when allocating space for a file/sharedfile instance.
85
- When disabling disk templates cluster-wide, the cluster now first
86
  checks whether there are instances still using those templates.
87
- 'gnt-node list-storage' now also reports storage information about
88
  file-based storage types.
89
- In case of non drbd instances, export \*_SECONDARY environment variables
90
  as empty strings (and not "None") during 'instance-migrate' related hooks.
91

    
92
New features
93
~~~~~~~~~~~~
94

    
95
- KVM hypervisors can now access RBD storage directly without having to
96
  go through a block device.
97
- A new command 'gnt-cluster upgrade' was added that automates the upgrade
98
  procedure between two Ganeti versions that are both 2.10 or higher.
99
- The move-instance command can now change disk templates when moving
100
  instances, and does not require any node placement options to be
101
  specified if the destination cluster has a default iallocator.
102
- Users can now change the soundhw and cpuid settings for XEN hypervisors.
103
- Hail and hbal now have the (optional) capability of accessing average CPU
104
  load information through the monitoring deamon, and to use it to dynamically
105
  adapt the allocation of instances.
106
- Hotplug support. Introduce new option '--hotplug' to ``gnt-instance modify``
107
  so that disk and NIC modifications take effect without the need of actual
108
  reboot. There are a couple of constrains currently for this feature:
109

    
110
   - only KVM hypervisor (versions >= 1.0) supports it,
111
   - one can not (yet) hotplug a disk using userspace access mode for RBD
112
   - in case of a downgrade instances should suffer a reboot in order to
113
     be migratable (due to core change of runtime files)
114

    
115
Misc changes
116
~~~~~~~~~~~~
117

    
118
- A new test framework for logical units was introduced and the test
119
  coverage for logical units was improved significantly.
120
- Opcodes are entirely generated from Haskell using the tool 'hs2py' and
121
  the module 'src/Ganeti/OpCodes.hs'.
122
- Constants are also generated from Haskell using the tool
123
  'hs2py-constants' and the module 'src/Ganeti/Constants.hs', with the
124
  exception of socket related constants, which require changing the
125
  cluster configuration file, and HVS related constants, because they
126
  are part of a port of instance queries to Haskell.  As a result, these
127
  changes will be part of the next release of Ganeti.
128

    
129
New dependencies
130
~~~~~~~~~~~~~~~~
131

    
132
The following new dependencies have been added/updated.
133

    
134
Python
135

    
136
- The version requirements for ``python-mock`` have increased to at least
137
  version 1.0.1. It is still used for testing only.
138

    
139
Since 2.10.0 beta1
140
~~~~~~~~~~~~~~~~~~
141

    
142
- All known issues in 2.10.0 beta1 have been resolved (see changes from
143
  the 2.8 branch).
144
- Improve handling of KVM runtime files from earlier Ganeti versions
145
- Documentation fixes
146

    
147
Inherited from the 2.9 branch:
148

    
149
- use custom KVM path if set for version checking
150
- SingleNotifyPipeCondition: don't share pollers
151

    
152
Inherited from the 2.8 branch:
153

    
154
- Fixed Luxi daemon socket permissions after master-failover
155
- Improve IP version detection code directly checking for colons rather than
156
  passing the family from the cluster object
157
- Fix NODE/NODE_RES locking in LUInstanceCreate by not acquiring NODE_RES locks
158
  opportunistically anymore (Issue 622)
159
- Allow link local IPv6 gateways (Issue 624)
160
- Fix error printing (Issue 616)
161
- Fix a bug in InstanceSetParams concerning names: in case no name is passed in
162
  disk modifications, keep the old one. If name=none then set disk name to
163
  None.
164
- Update build_chroot script to work with the latest hackage packages
165
- Add a packet number limit to "fping" in master-ip-setup (Issue 630)
166
- Fix evacuation out of drained node (Issue 615)
167
- Add default file_driver if missing (Issue 571)
168
- Fix job error message after unclean master shutdown (Issue 618)
169
- Lock group(s) when creating instances (Issue 621)
170
- SetDiskID() before accepting an instance (Issue 633)
171
- Allow the ext template disks to receive arbitrary parameters, both at creation
172
  time and while being modified
173
- Xen handle domain shutdown (future proofing cherry-pick)
174
- Refactor reading live data in htools (future proofing cherry-pick)
175

    
176

    
177
Version 2.10.0 beta1
178
--------------------
179

    
180
*(Released Wed, 27 Nov 2013)*
181

    
182
This was the first beta release of the 2.10 series. All important changes
183
are listed in the latest 2.10 entry.
184

    
185
Known issues
186
~~~~~~~~~~~~
187

    
188
The following issues are known to be present in the beta and will be fixed
189
before rc1.
190

    
191
- Issue 477: Wrong permissions for confd LUXI socket
192
- Issue 621: Instance related opcodes do not aquire network/group locks
193
- Issue 622: Assertion Error: Node locks differ from node resource locks
194
- Issue 623: IPv6 Masterd <-> Luxid communication error
195

    
196

    
197
Version 2.9.3
198
-------------
199

    
200
*(Released Mon, 27 Jan 2014)*
201

    
202
- Ensure that all the hypervisors exist in the config file (Issue 640)
203
- Correctly recognise the role as master node (Issue 687)
204
- configure: allow detection of Sphinx 1.2+ (Issue 502)
205
- gnt-instance now honors the KVM path correctly (Issue 691)
206

    
207
Inherited from the 2.8 branch:
208

    
209
- Change the list separator for the usb_devices parameter from comma to space.
210
  Commas could not work because they are already the hypervisor option
211
  separator (Issue 649)
212
- Add support for blktap2 file-driver (Issue 638)
213
- Add network tag definitions to the haskell codebase (Issue 641)
214
- Fix RAPI network tag handling
215
- Add the network tags to the tags searched by gnt-cluster search-tags
216
- Fix caching bug preventing jobs from being cancelled
217
- Start-master/stop-master was always failing if ConfD was disabled. (Issue 685)
218

    
219

    
220
Version 2.9.2
221
-------------
222

    
223
*(Released Fri, 13 Dec 2013)*
224

    
225
- use custom KVM path if set for version checking
226
- SingleNotifyPipeCondition: don't share pollers
227

    
228
Inherited from the 2.8 branch:
229

    
230
- Fixed Luxi daemon socket permissions after master-failover
231
- Improve IP version detection code directly checking for colons rather than
232
  passing the family from the cluster object
233
- Fix NODE/NODE_RES locking in LUInstanceCreate by not acquiring NODE_RES locks
234
  opportunistically anymore (Issue 622)
235
- Allow link local IPv6 gateways (Issue 624)
236
- Fix error printing (Issue 616)
237
- Fix a bug in InstanceSetParams concerning names: in case no name is passed in
238
  disk modifications, keep the old one. If name=none then set disk name to
239
  None.
240
- Update build_chroot script to work with the latest hackage packages
241
- Add a packet number limit to "fping" in master-ip-setup (Issue 630)
242
- Fix evacuation out of drained node (Issue 615)
243
- Add default file_driver if missing (Issue 571)
244
- Fix job error message after unclean master shutdown (Issue 618)
245
- Lock group(s) when creating instances (Issue 621)
246
- SetDiskID() before accepting an instance (Issue 633)
247
- Allow the ext template disks to receive arbitrary parameters, both at creation
248
  time and while being modified
249
- Xen handle domain shutdown (future proofing cherry-pick)
250
- Refactor reading live data in htools (future proofing cherry-pick)
251

    
252

    
253
Version 2.9.1
254
-------------
255

    
256
*(Released Wed, 13 Nov 2013)*
257

    
258
- fix bug, that kept nodes offline when readding
259
- when verifying DRBD versions, ignore unavailable nodes
260
- fix bug that made the console unavailable on kvm in split-user
261
  setup (issue 608)
262
- DRBD: ensure peers are UpToDate for dual-primary (inherited 2.8.2)
263

    
264

    
265
Version 2.9.0
266
-------------
267

    
268
*(Released Tue, 5 Nov 2013)*
269

    
270
Incompatible/important changes
271
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
272

    
273
- hroller now also plans for capacity to move non-redundant instances off
274
  any node to be rebooted; the old behavior of completely ignoring any
275
  non-redundant instances can be restored by adding the --ignore-non-redundant
276
  option.
277
- The cluster option '--no-lvm-storage' was removed in favor of the new option
278
  '--enabled-disk-templates'.
279
- On instance creation, disk templates no longer need to be specified
280
  with '-t'. The default disk template will be taken from the list of
281
  enabled disk templates.
282
- The monitoring daemon is now running as root, in order to be able to collect
283
  information only available to root (such as the state of Xen instances).
284
- The ConfD client is now IPv6 compatible.
285
- File and shared file storage is no longer dis/enabled at configure time,
286
  but using the option '--enabled-disk-templates' at cluster initialization and
287
  modification.
288
- The default directories for file and shared file storage are not anymore
289
  specified at configure time, but taken from the cluster's configuration.
290
  They can be set at cluster initialization and modification with
291
  '--file-storage-dir' and '--shared-file-storage-dir'.
292
- Cluster verification now includes stricter checks regarding the
293
  default file and shared file storage directories. It now checks that
294
  the directories are explicitely allowed in the 'file-storage-paths' file and
295
  that the directories exist on all nodes.
296
- The list of allowed disk templates in the instance policy and the list
297
  of cluster-wide enabled disk templates is now checked for consistency
298
  on cluster or group modification. On cluster initialization, the ipolicy
299
  disk templates are ensured to be a subset of the cluster-wide enabled
300
  disk templates.
301

    
302
New features
303
~~~~~~~~~~~~
304

    
305
- DRBD 8.4 support. Depending on the installed DRBD version, Ganeti now uses
306
  the correct command syntax. It is possible to use different DRBD versions
307
  on different nodes as long as they are compatible to each other. This
308
  enables rolling upgrades of DRBD with no downtime. As permanent operation
309
  of different DRBD versions within a node group is discouraged,
310
  ``gnt-cluster verify`` will emit a warning if it detects such a situation.
311
- New "inst-status-xen" data collector for the monitoring daemon, providing
312
  information about the state of the xen instances on the nodes.
313
- New "lv" data collector for the monitoring daemon, collecting data about the
314
  logical volumes on the nodes, and pairing them with the name of the instances
315
  they belong to.
316
- New "diskstats" data collector, collecting the data from /proc/diskstats and
317
  presenting them over the monitoring daemon interface.
318
- The ConfD client is now IPv6 compatible.
319

    
320
New dependencies
321
~~~~~~~~~~~~~~~~
322
The following new dependencies have been added.
323

    
324
Python
325

    
326
- ``python-mock`` (http://www.voidspace.org.uk/python/mock/) is now a required
327
  for the unit tests (and only used for testing).
328

    
329
Haskell
330

    
331
- ``hslogger`` (http://software.complete.org/hslogger) is now always
332
  required, even if confd is not enabled.
333

    
334
Since 2.9.0 rc3
335
~~~~~~~~~~~~~~~
336

    
337
- Correctly start/stop luxid during gnt-cluster master-failover (inherited
338
  from stable-2.8)
339
- Improved error messsages (inherited from stable-2.8)
340

    
341

    
342
Version 2.9.0 rc3
343
-----------------
344

    
345
*(Released Tue, 15 Oct 2013)*
346

    
347
The third release candidate in the 2.9 series. Since 2.9.0 rc2:
348

    
349
- in implicit configuration upgrade, match ipolicy with enabled disk templates
350
- improved harep documentation (inherited from stable-2.8)
351

    
352

    
353
Version 2.9.0 rc2
354
-----------------
355

    
356
*(Released Wed, 9 Oct 2013)*
357

    
358
The second release candidate in the 2.9 series. Since 2.9.0 rc1:
359

    
360
- Fix bug in cfgupgrade that led to failure when upgrading from 2.8 with
361
  at least one DRBD instance.
362
- Fix bug in cfgupgrade that led to an invalid 2.8 configuration after
363
  downgrading.
364

    
365

    
366
Version 2.9.0 rc1
367
-----------------
368

    
369
*(Released Tue, 1 Oct 2013)*
370

    
371
The first release candidate in the 2.9 series. Since 2.9.0 beta1:
372

    
373
- various bug fixes
374
- update of the documentation, in particular installation instructions
375
- merging of LD_* constants into DT_* constants
376
- python style changes to be compatible with newer versions of pylint
377

    
378

    
379
Version 2.9.0 beta1
380
-------------------
381

    
382
*(Released Thu, 29 Aug 2013)*
383

    
384
This was the first beta release of the 2.9 series. All important changes
385
are listed in the latest 2.9 entry.
386

    
387

    
388
Version 2.8.4
389
-------------
390

    
391
*(Released Thu, 23 Jan 2014)*
392

    
393
- Change the list separator for the usb_devices parameter from comma to space.
394
  Commas could not work because they are already the hypervisor option
395
  separator (Issue 649)
396
- Add support for blktap2 file-driver (Issue 638)
397
- Add network tag definitions to the haskell codebase (Issue 641)
398
- Fix RAPI network tag handling
399
- Add the network tags to the tags searched by gnt-cluster search-tags
400
- Fix caching bug preventing jobs from being cancelled
401
- Start-master/stop-master was always failing if ConfD was disabled. (Issue 685)
402

    
403

    
404
Version 2.8.3
405
-------------
406

    
407
*(Released Thu, 12 Dec 2013)*
408

    
409
- Fixed Luxi daemon socket permissions after master-failover
410
- Improve IP version detection code directly checking for colons rather than
411
  passing the family from the cluster object
412
- Fix NODE/NODE_RES locking in LUInstanceCreate by not acquiring NODE_RES locks
413
  opportunistically anymore (Issue 622)
414
- Allow link local IPv6 gateways (Issue 624)
415
- Fix error printing (Issue 616)
416
- Fix a bug in InstanceSetParams concerning names: in case no name is passed in
417
  disk modifications, keep the old one. If name=none then set disk name to
418
  None.
419
- Update build_chroot script to work with the latest hackage packages
420
- Add a packet number limit to "fping" in master-ip-setup (Issue 630)
421
- Fix evacuation out of drained node (Issue 615)
422
- Add default file_driver if missing (Issue 571)
423
- Fix job error message after unclean master shutdown (Issue 618)
424
- Lock group(s) when creating instances (Issue 621)
425
- SetDiskID() before accepting an instance (Issue 633)
426
- Allow the ext template disks to receive arbitrary parameters, both at creation
427
  time and while being modified
428
- Xen handle domain shutdown (future proofing cherry-pick)
429
- Refactor reading live data in htools (future proofing cherry-pick)
430

    
431

    
432
Version 2.8.2
433
-------------
434

    
435
*(Released Thu, 07 Nov 2013)*
436

    
437
- DRBD: ensure peers are UpToDate for dual-primary
438
- Improve error message for replace-disks
439
- More dependency checks at configure time
440
- Placate warnings on ganeti.outils_unittest.py
441

    
442

    
443
Version 2.8.1
444
-------------
445

    
446
*(Released Thu, 17 Oct 2013)*
447

    
448
- Correctly start/stop luxid during gnt-cluster master-failover
449
- Don't attempt IPv6 ssh in case of IPv4 cluster (Issue 595)
450
- Fix path for the job queue serial file
451
- Improved harep man page
452
- Minor documentation improvements
453

    
454

    
455
Version 2.8.0
456
-------------
457

    
458
*(Released Mon, 30 Sep 2013)*
459

    
460
Incompatible/important changes
461
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
462

    
463
- Instance policy can contain multiple instance specs, as described in
464
  the โ€œConstrained instance sizesโ€ section of :doc:`Partitioned Ganeti
465
  <design-partitioned>`. As a consequence, it's not possible to partially change
466
  or override instance specs. Bounding specs (min and max) can be specified as a
467
  whole using the new option ``--ipolicy-bounds-specs``, while standard
468
  specs use the new option ``--ipolicy-std-specs``.
469
- The output of the info command of gnt-cluster, gnt-group, gnt-node,
470
  gnt-instance is a valid YAML object.
471
- hail now honors network restrictions when allocating nodes. This led to an
472
  update of the IAllocator protocol. See the IAllocator documentation for
473
  details.
474
- confd now only answers static configuration request over the network. luxid
475
  was extracted, listens on the local LUXI socket and responds to live queries.
476
  This allows finer grained permissions if using separate users.
477

    
478
New features
479
~~~~~~~~~~~~
480

    
481
- The :doc:`Remote API <rapi>` daemon now supports a command line flag
482
  to always require authentication, ``--require-authentication``. It can
483
  be specified in ``$sysconfdir/default/ganeti``.
484
- A new cluster attribute 'enabled_disk_templates' is introduced. It will
485
  be used to manage the disk templates to be used by instances in the cluster.
486
  Initially, it will be set to a list that includes plain, drbd, if they were
487
  enabled by specifying a volume group name, and file and sharedfile, if those
488
  were enabled at configure time. Additionally, it will include all disk
489
  templates that are currently used by instances. The order of disk templates
490
  will be based on Ganeti's history of supporting them. In the future, the
491
  first entry of the list will be used as a default disk template on instance
492
  creation.
493
- ``cfgupgrade`` now supports a ``--downgrade`` option to bring the
494
  configuration back to the previous stable version.
495
- Disk templates in group ipolicy can be restored to the default value.
496
- Initial support for diskless instances and virtual clusters in QA.
497
- More QA and unit tests for instance policies.
498
- Every opcode now contains a reason trail (visible through ``gnt-job info``)
499
  describing why the opcode itself was executed.
500
- The monitoring daemon is now available. It allows users to query the cluster
501
  for obtaining information about the status of the system. The daemon is only
502
  responsible for providing the information over the network: the actual data
503
  gathering is performed by data collectors (currently, only the DRBD status
504
  collector is available).
505
- In order to help developers work on Ganeti, a new script
506
  (``devel/build_chroot``) is provided, for building a chroot that contains all
507
  the required development libraries and tools for compiling Ganeti on a Debian
508
  Squeeze system.
509
- A new tool, ``harep``, for performing self-repair and recreation of instances
510
  in Ganeti has been added.
511
- Split queries are enabled for tags, network, exports, cluster info, groups,
512
  jobs, nodes.
513
- New command ``show-ispecs-cmd`` for ``gnt-cluster`` and ``gnt-group``.
514
  It prints the command line to set the current policies, to ease
515
  changing them.
516
- Add the ``vnet_hdr`` HV parameter for KVM, to control whether the tap
517
  devices for KVM virtio-net interfaces will get created with VNET_HDR
518
  (IFF_VNET_HDR) support. If set to false, it disables offloading on the
519
  virtio-net interfaces, which prevents host kernel tainting and log
520
  flooding, when dealing with broken or malicious virtio-net drivers.
521
  It's set to true by default.
522
- Instance failover now supports a ``--cleanup`` parameter for fixing previous
523
  failures.
524
- Support 'viridian' parameter in Xen HVM
525
- Support DSA SSH keys in bootstrap
526
- To simplify the work of packaging frameworks that want to add the needed users
527
  and groups in a split-user setup themselves, at build time three files in
528
  ``doc/users`` will be generated. The ``groups`` files contains, one per line,
529
  the groups to be generated, the ``users`` file contains, one per line, the
530
  users to be generated, optionally followed by their primary group, where
531
  important. The ``groupmemberships`` file contains, one per line, additional
532
  user-group membership relations that need to be established. The syntax of
533
  these files will remain stable in all future versions.
534

    
535

    
536
New dependencies
537
~~~~~~~~~~~~~~~~
538
The following new dependencies have been added:
539

    
540
For Haskell:
541
- The ``curl`` library is not optional anymore for compiling the Haskell code.
542
- ``snap-server`` library (if monitoring is enabled).
543

    
544
For Python:
545
- The minimum Python version needed to run Ganeti is now 2.6.
546
- ``yaml`` library (only for running the QA).
547

    
548
Since 2.8.0 rc3
549
~~~~~~~~~~~~~~~
550
- Perform proper cleanup on termination of Haskell daemons
551
- Fix corner-case in handling of remaining retry time
552

    
553

    
554
Version 2.8.0 rc3
555
-----------------
556

    
557
*(Released Tue, 17 Sep 2013)*
558

    
559
- To simplify the work of packaging frameworks that want to add the needed users
560
  and groups in a split-user setup themselves, at build time three files in
561
  ``doc/users`` will be generated. The ``groups`` files contains, one per line,
562
  the groups to be generated, the ``users`` file contains, one per line, the
563
  users to be generated, optionally followed by their primary group, where
564
  important. The ``groupmemberships`` file contains, one per line, additional
565
  user-group membership relations that need to be established. The syntax of
566
  these files will remain stable in all future versions.
567
- Add a default to file-driver when unspecified over RAPI (Issue 571)
568
- Mark the DSA host pubkey as optional, and remove it during config downgrade
569
  (Issue 560)
570
- Some documentation fixes
571

    
572

    
573
Version 2.8.0 rc2
574
-----------------
575

    
576
*(Released Tue, 27 Aug 2013)*
577

    
578
The second release candidate of the 2.8 series. Since 2.8.0. rc1:
579

    
580
- Support 'viridian' parameter in Xen HVM (Issue 233)
581
- Include VCS version in ``gnt-cluster version``
582
- Support DSA SSH keys in bootstrap (Issue 338)
583
- Fix batch creation of instances
584
- Use FQDN to check master node status (Issue 551)
585
- Make the DRBD collector more failure-resilient
586

    
587

    
588
Version 2.8.0 rc1
589
-----------------
590

    
591
*(Released Fri, 2 Aug 2013)*
592

    
593
The first release candidate of the 2.8 series. Since 2.8.0 beta1:
594

    
595
- Fix upgrading/downgrading from 2.7
596
- Increase maximum RAPI message size
597
- Documentation updates
598
- Split ``confd`` between ``luxid`` and ``confd``
599
- Merge 2.7 series up to the 2.7.1 release
600
- Allow the ``modify_etc_hosts`` option to be changed
601
- Add better debugging for ``luxid`` queries
602
- Expose bulk parameter for GetJobs in RAPI client
603
- Expose missing ``network`` fields in RAPI
604
- Add some ``cluster verify`` tests
605
- Some unittest fixes
606
- Fix a malfunction in ``hspace``'s tiered allocation
607
- Fix query compatibility between haskell and python implementations
608
- Add the ``vnet_hdr`` HV parameter for KVM
609
- Add ``--cleanup`` to instance failover
610
- Change the connected groups format in ``gnt-network info`` output; it
611
  was previously displayed as a raw list by mistake. (Merged from 2.7)
612

    
613

    
614
Version 2.8.0 beta1
615
-------------------
616

    
617
*(Released Mon, 24 Jun 2013)*
618

    
619
This was the first beta release of the 2.8 series. All important changes
620
are listed in the latest 2.8 entry.
621

    
622

    
623
Version 2.7.2
624
-------------
625

    
626
*(Released Thu, 26 Sep 2013)*
627

    
628
- Change the connected groups format in ``gnt-network info`` output; it
629
  was previously displayed as a raw list by mistake
630
- Check disk template in right dict when copying
631
- Support multi-instance allocs without iallocator
632
- Fix some errors in the documentation
633
- Fix formatting of tuple in an error message
634

    
635

    
636
Version 2.7.1
637
-------------
638

    
639
*(Released Thu, 25 Jul 2013)*
640

    
641
- Add logrotate functionality in daemon-util
642
- Add logrotate example file
643
- Add missing fields to network queries over rapi
644
- Fix network object timestamps
645
- Add support for querying network timestamps
646
- Fix a typo in the example crontab
647
- Fix a documentation typo
648

    
649

    
650
Version 2.7.0
651
-------------
652

    
653
*(Released Thu, 04 Jul 2013)*
654

    
655
Incompatible/important changes
656
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
657

    
658
- Instance policies for disk size were documented to be on a per-disk
659
  basis, but hail applied them to the sum of all disks. This has been
660
  fixed.
661
- ``hbal`` will now exit with status 0 if, during job execution over
662
  LUXI, early exit has been requested and all jobs are successful;
663
  before, exit status 1 was used, which cannot be differentiated from
664
  "job error" case
665
- Compatibility with newer versions of rbd has been fixed
666
- ``gnt-instance batch-create`` has been changed to use the bulk create
667
  opcode from Ganeti. This lead to incompatible changes in the format of
668
  the JSON file. It's now not a custom dict anymore but a dict
669
  compatible with the ``OpInstanceCreate`` opcode.
670
- Parent directories for file storage need to be listed in
671
  ``$sysconfdir/ganeti/file-storage-paths`` now. ``cfgupgrade`` will
672
  write the file automatically based on old configuration values, but it
673
  can not distribute it across all nodes and the file contents should be
674
  verified. Use ``gnt-cluster copyfile
675
  $sysconfdir/ganeti/file-storage-paths`` once the cluster has been
676
  upgraded. The reason for requiring this list of paths now is that
677
  before it would have been possible to inject new paths via RPC,
678
  allowing files to be created in arbitrary locations. The RPC protocol
679
  is protected using SSL/X.509 certificates, but as a design principle
680
  Ganeti does not permit arbitrary paths to be passed.
681
- The parsing of the variants file for OSes (see
682
  :manpage:`ganeti-os-interface(7)`) has been slightly changed: now empty
683
  lines and comment lines (starting with ``#``) are ignored for better
684
  readability.
685
- The ``setup-ssh`` tool added in Ganeti 2.2 has been replaced and is no
686
  longer available. ``gnt-node add`` now invokes a new tool on the
687
  destination node, named ``prepare-node-join``, to configure the SSH
688
  daemon. Paramiko is no longer necessary to configure nodes' SSH
689
  daemons via ``gnt-node add``.
690
- Draining (``gnt-cluster queue drain``) and un-draining the job queue
691
  (``gnt-cluster queue undrain``) now affects all nodes in a cluster and
692
  the flag is not reset after a master failover.
693
- Python 2.4 has *not* been tested with this release. Using 2.6 or above
694
  is recommended. 2.6 will be mandatory from the 2.8 series.
695

    
696

    
697
New features
698
~~~~~~~~~~~~
699

    
700
- New network management functionality to support automatic allocation
701
  of IP addresses and managing of network parameters. See
702
  :manpage:`gnt-network(8)` for more details.
703
- New external storage backend, to allow managing arbitrary storage
704
  systems external to the cluster. See
705
  :manpage:`ganeti-extstorage-interface(7)`.
706
- New ``exclusive-storage`` node parameter added, restricted to
707
  nodegroup level. When it's set to true, physical disks are assigned in
708
  an exclusive fashion to instances, as documented in :doc:`Partitioned
709
  Ganeti <design-partitioned>`.  Currently, only instances using the
710
  ``plain`` disk template are supported.
711
- The KVM hypervisor has been updated with many new hypervisor
712
  parameters, including a generic one for passing arbitrary command line
713
  values. See a complete list in :manpage:`gnt-instance(8)`. It is now
714
  compatible up to qemu 1.4.
715
- A new tool, called ``mon-collector``, is the stand-alone executor of
716
  the data collectors for a monitoring system. As of this version, it
717
  just includes the DRBD data collector, that can be executed by calling
718
  ``mon-collector`` using the ``drbd`` parameter. See
719
  :manpage:`mon-collector(7)`.
720
- A new user option, :pyeval:`rapi.RAPI_ACCESS_READ`, has been added
721
  for RAPI users. It allows granting permissions to query for
722
  information to a specific user without giving
723
  :pyeval:`rapi.RAPI_ACCESS_WRITE` permissions.
724
- A new tool named ``node-cleanup`` has been added. It cleans remains of
725
  a cluster from a machine by stopping all daemons, removing
726
  certificates and ssconf files. Unless the ``--no-backup`` option is
727
  given, copies of the certificates are made.
728
- Instance creations now support the use of opportunistic locking,
729
  potentially speeding up the (parallel) creation of multiple instances.
730
  This feature is currently only available via the :doc:`RAPI
731
  <rapi>` interface and when an instance allocator is used. If the
732
  ``opportunistic_locking`` parameter is set the opcode will try to
733
  acquire as many locks as possible, but will not wait for any locks
734
  held by other opcodes. If not enough resources can be found to
735
  allocate the instance, the temporary error code
736
  :pyeval:`errors.ECODE_TEMP_NORES` is returned. The operation can be
737
  retried thereafter, with or without opportunistic locking.
738
- New experimental linux-ha resource scripts.
739
- Restricted-commands support: ganeti can now be asked (via command line
740
  or rapi) to perform commands on a node. These are passed via ganeti
741
  RPC rather than ssh. This functionality is restricted to commands
742
  specified on the ``$sysconfdir/ganeti/restricted-commands`` for security
743
  reasons. The file is not copied automatically.
744

    
745

    
746
Misc changes
747
~~~~~~~~~~~~
748

    
749
- Diskless instances are now externally mirrored (Issue 237). This for
750
  now has only been tested in conjunction with explicit target nodes for
751
  migration/failover.
752
- Queries not needing locks or RPC access to the node can now be
753
  performed by the confd daemon, making them independent from jobs, and
754
  thus faster to execute. This is selectable at configure time.
755
- The functionality for allocating multiple instances at once has been
756
  overhauled and is now also available through :doc:`RAPI <rapi>`.
757

    
758
There are no significant changes from version 2.7.0~rc3.
759

    
760

    
761
Version 2.7.0 rc3
762
-----------------
763

    
764
*(Released Tue, 25 Jun 2013)*
765

    
766
- Fix permissions on the confd query socket (Issue 477)
767
- Fix permissions on the job archive dir (Issue 498)
768
- Fix handling of an internal exception in replace-disks (Issue 472)
769
- Fix gnt-node info handling of shortened names (Issue 497)
770
- Fix gnt-instance grow-disk when wiping is enabled
771
- Documentation improvements, and support for newer pandoc
772
- Fix hspace honoring ipolicy for disks (Issue 484)
773
- Improve handling of the ``kvm_extra`` HV parameter
774

    
775

    
776
Version 2.7.0 rc2
777
-----------------
778

    
779
*(Released Fri, 24 May 2013)*
780

    
781
- ``devel/upload`` now works when ``/var/run`` on the target nodes is a
782
  symlink.
783
- Disks added through ``gnt-instance modify`` or created through
784
  ``gnt-instance recreate-disks`` are wiped, if the
785
  ``prealloc_wipe_disks`` flag is set.
786
- If wiping newly created disks fails, the disks are removed. Also,
787
  partial failures in creating disks through ``gnt-instance modify``
788
  triggers a cleanup of the partially-created disks.
789
- Removing the master IP address doesn't fail if the address has been
790
  already removed.
791
- Fix ownership of the OS log dir
792
- Workaround missing SO_PEERCRED constant (Issue 191)
793

    
794

    
795
Version 2.7.0 rc1
796
-----------------
797

    
798
*(Released Fri, 3 May 2013)*
799

    
800
This was the first release candidate of the 2.7 series. Since beta3:
801

    
802
- Fix kvm compatibility with qemu 1.4 (Issue 389)
803
- Documentation updates (admin guide, upgrade notes, install
804
  instructions) (Issue 372)
805
- Fix gnt-group list nodes and instances count (Issue 436)
806
- Fix compilation without non-mandatory libraries (Issue 441)
807
- Fix xen-hvm hypervisor forcing nics to type 'ioemu' (Issue 247)
808
- Make confd logging more verbose at INFO level (Issue 435)
809
- Improve "networks" documentation in :manpage:`gnt-instance(8)`
810
- Fix failure path for instance storage type conversion (Issue 229)
811
- Update htools text backend documentation
812
- Improve the renew-crypto section of :manpage:`gnt-cluster(8)`
813
- Disable inter-cluster instance move for file-based instances, because
814
  it is dependant on instance export, which is not supported for
815
  file-based instances. (Issue 414)
816
- Fix gnt-job crashes on non-ascii characters (Issue 427)
817
- Fix volume group checks on non-vm-capable nodes (Issue 432)
818

    
819

    
820
Version 2.7.0 beta3
821
-------------------
822

    
823
*(Released Mon, 22 Apr 2013)*
824

    
825
This was the third beta release of the 2.7 series. Since beta2:
826

    
827
- Fix hail to verify disk instance policies on a per-disk basis (Issue 418).
828
- Fix data loss on wrong usage of ``gnt-instance move``
829
- Properly export errors in confd-based job queries
830
- Add ``users-setup`` tool
831
- Fix iallocator protocol to report 0 as a disk size for diskless
832
  instances. This avoids hail breaking when a diskless instance is
833
  present.
834
- Fix job queue directory permission problem that made confd job queries
835
  fail. This requires running an ``ensure-dirs --full-run`` on upgrade
836
  for access to archived jobs (Issue 406).
837
- Limit the sizes of networks supported by ``gnt-network`` to something
838
  between a ``/16`` and a ``/30`` to prevent memory bloat and crashes.
839
- Fix bugs in instance disk template conversion
840
- Fix GHC 7 compatibility
841
- Fix ``burnin`` install path (Issue 426).
842
- Allow very small disk grows (Issue 347).
843
- Fix a ``ganeti-noded`` memory bloat introduced in 2.5, by making sure
844
  that noded doesn't import masterd code (Issue 419).
845
- Make sure the default metavg at cluster init is the same as the vg, if
846
  unspecified (Issue 358).
847
- Fix cleanup of partially created disks (part of Issue 416)
848

    
849

    
850
Version 2.7.0 beta2
851
-------------------
852

    
853
*(Released Tue, 2 Apr 2013)*
854

    
855
This was the second beta release of the 2.7 series. Since beta1:
856

    
857
- Networks no longer have a "type" slot, since this information was
858
  unused in Ganeti: instead of it tags should be used.
859
- The rapi client now has a ``target_node`` option to MigrateInstance.
860
- Fix early exit return code for hbal (Issue 386).
861
- Fix ``gnt-instance migrate/failover -n`` (Issue 396).
862
- Fix ``rbd showmapped`` output parsing (Issue 312).
863
- Networks are now referenced indexed by UUID, rather than name. This
864
  will require running cfgupgrade, from 2.7.0beta1, if networks are in
865
  use.
866
- The OS environment now includes network information.
867
- Deleting of a network is now disallowed if any instance nic is using
868
  it, to prevent dangling references.
869
- External storage is now documented in man pages.
870
- The exclusive_storage flag can now only be set at nodegroup level.
871
- Hbal can now submit an explicit priority with its jobs.
872
- Many network related locking fixes.
873
- Bump up the required pylint version to 0.25.1.
874
- Fix the ``no_remember`` option in RAPI client.
875
- Many ipolicy related tests, qa, and fixes.
876
- Many documentation improvements and fixes.
877
- Fix building with ``--disable-file-storage``.
878
- Fix ``-q`` option in htools, which was broken if passed more than
879
  once.
880
- Some haskell/python interaction improvements and fixes.
881
- Fix iallocator in case of missing LVM storage.
882
- Fix confd config load in case of ``--no-lvm-storage``.
883
- The confd/query functionality is now mentioned in the security
884
  documentation.
885

    
886

    
887
Version 2.7.0 beta1
888
-------------------
889

    
890
*(Released Wed, 6 Feb 2013)*
891

    
892
This was the first beta release of the 2.7 series. All important changes
893
are listed in the latest 2.7 entry.
894

    
895

    
896
Version 2.6.2
897
-------------
898

    
899
*(Released Fri, 21 Dec 2012)*
900

    
901
Important behaviour change: hbal won't rebalance anymore instances which
902
have the ``auto_balance`` attribute set to false. This was the intention
903
all along, but until now it only skipped those from the N+1 memory
904
reservation (DRBD-specific).
905

    
906
A significant number of bug fixes in this release:
907

    
908
- Fixed disk adoption interaction with ipolicy checks.
909
- Fixed networking issues when instances are started, stopped or
910
  migrated, by forcing the tap device's MAC prefix to "fe" (issue 217).
911
- Fixed the warning in cluster verify for shared storage instances not
912
  being redundant.
913
- Fixed removal of storage directory on shared file storage (issue 262).
914
- Fixed validation of LVM volume group name in OpClusterSetParams
915
  (``gnt-cluster modify``) (issue 285).
916
- Fixed runtime memory increases (``gnt-instance modify -m``).
917
- Fixed live migration under Xen's ``xl`` mode.
918
- Fixed ``gnt-instance console`` with ``xl``.
919
- Fixed building with newer Haskell compiler/libraries.
920
- Fixed PID file writing in Haskell daemons (confd); this prevents
921
  restart issues if confd was launched manually (outside of
922
  ``daemon-util``) while another copy of it was running
923
- Fixed a type error when doing live migrations with KVM (issue 297) and
924
  the error messages for failing migrations have been improved.
925
- Fixed opcode validation for the out-of-band commands (``gnt-node
926
  power``).
927
- Fixed a type error when unsetting OS hypervisor parameters (issue
928
  311); now it's possible to unset all OS-specific hypervisor
929
  parameters.
930
- Fixed the ``dry-run`` mode for many operations: verification of
931
  results was over-zealous but didn't take into account the ``dry-run``
932
  operation, resulting in "wrong" failures.
933
- Fixed bash completion in ``gnt-job list`` when the job queue has
934
  hundreds of entries; especially with older ``bash`` versions, this
935
  results in significant CPU usage.
936

    
937
And lastly, a few other improvements have been made:
938

    
939
- Added option to force master-failover without voting (issue 282).
940
- Clarified error message on lock conflict (issue 287).
941
- Logging of newly submitted jobs has been improved (issue 290).
942
- Hostname checks have been made uniform between instance rename and
943
  create (issue 291).
944
- The ``--submit`` option is now supported by ``gnt-debug delay``.
945
- Shutting down the master daemon by sending SIGTERM now stops it from
946
  processing jobs waiting for locks; instead, those jobs will be started
947
  once again after the master daemon is started the next time (issue
948
  296).
949
- Support for Xen's ``xl`` program has been improved (besides the fixes
950
  above).
951
- Reduced logging noise in the Haskell confd daemon (only show one log
952
  entry for each config reload, instead of two).
953
- Several man page updates and typo fixes.
954

    
955

    
956
Version 2.6.1
957
-------------
958

    
959
*(Released Fri, 12 Oct 2012)*
960

    
961
A small bugfix release. Among the bugs fixed:
962

    
963
- Fixed double use of ``PRIORITY_OPT`` in ``gnt-node migrate``, that
964
  made the command unusable.
965
- Commands that issue many jobs don't fail anymore just because some jobs
966
  take so long that other jobs are archived.
967
- Failures during ``gnt-instance reinstall`` are reflected by the exit
968
  status.
969
- Issue 190 fixed. Check for DRBD in cluster verify is enabled only when
970
  DRBD is enabled.
971
- When ``always_failover`` is set, ``--allow-failover`` is not required
972
  in migrate commands anymore.
973
- ``bash_completion`` works even if extglob is disabled.
974
- Fixed bug with locks that made failover for RDB-based instances fail.
975
- Fixed bug in non-mirrored instance allocation that made Ganeti choose
976
  a random node instead of one based on the allocator metric.
977
- Support for newer versions of pylint and pep8.
978
- Hail doesn't fail anymore when trying to add an instance of type
979
  ``file``, ``sharedfile`` or ``rbd``.
980
- Added new Makefile target to rebuild the whole distribution, so that
981
  all files are included.
982

    
983

    
984
Version 2.6.0
985
-------------
986

    
987
*(Released Fri, 27 Jul 2012)*
988

    
989

    
990
.. attention:: The ``LUXI`` protocol has been made more consistent
991
   regarding its handling of command arguments. This, however, leads to
992
   incompatibility issues with previous versions. Please ensure that you
993
   restart Ganeti daemons soon after the upgrade, otherwise most
994
   ``LUXI`` calls (job submission, setting/resetting the drain flag,
995
   pausing/resuming the watcher, cancelling and archiving jobs, querying
996
   the cluster configuration) will fail.
997

    
998

    
999
New features
1000
~~~~~~~~~~~~
1001

    
1002
Instance run status
1003
+++++++++++++++++++
1004

    
1005
The current ``admin_up`` field, which used to denote whether an instance
1006
should be running or not, has been removed. Instead, ``admin_state`` is
1007
introduced, with 3 possible values -- ``up``, ``down`` and ``offline``.
1008

    
1009
The rational behind this is that an instance being โ€œdownโ€ can have
1010
different meanings:
1011

    
1012
- it could be down during a reboot
1013
- it could be temporarily be down for a reinstall
1014
- or it could be down because it is deprecated and kept just for its
1015
  disk
1016

    
1017
The previous Boolean state was making it difficult to do capacity
1018
calculations: should Ganeti reserve memory for a down instance? Now, the
1019
tri-state field makes it clear:
1020

    
1021
- in ``up`` and ``down`` state, all resources are reserved for the
1022
  instance, and it can be at any time brought up if it is down
1023
- in ``offline`` state, only disk space is reserved for it, but not
1024
  memory or CPUs
1025

    
1026
The field can have an extra use: since the transition between ``up`` and
1027
``down`` and vice-versus is done via ``gnt-instance start/stop``, but
1028
transition between ``offline`` and ``down`` is done via ``gnt-instance
1029
modify``, it is possible to given different rights to users. For
1030
example, owners of an instance could be allowed to start/stop it, but
1031
not transition it out of the offline state.
1032

    
1033
Instance policies and specs
1034
+++++++++++++++++++++++++++
1035

    
1036
In previous Ganeti versions, an instance creation request was not
1037
limited on the minimum size and on the maximum size just by the cluster
1038
resources. As such, any policy could be implemented only in third-party
1039
clients (RAPI clients, or shell wrappers over ``gnt-*``
1040
tools). Furthermore, calculating cluster capacity via ``hspace`` again
1041
required external input with regards to instance sizes.
1042

    
1043
In order to improve these workflows and to allow for example better
1044
per-node group differentiation, we introduced instance specs, which
1045
allow declaring:
1046

    
1047
- minimum instance disk size, disk count, memory size, cpu count
1048
- maximum values for the above metrics
1049
- and โ€œstandardโ€ values (used in ``hspace`` to calculate the standard
1050
  sized instances)
1051

    
1052
The minimum/maximum values can be also customised at node-group level,
1053
for example allowing more powerful hardware to support bigger instance
1054
memory sizes.
1055

    
1056
Beside the instance specs, there are a few other settings belonging to
1057
the instance policy framework. It is possible now to customise, per
1058
cluster and node-group:
1059

    
1060
- the list of allowed disk templates
1061
- the maximum ratio of VCPUs per PCPUs (to control CPU oversubscription)
1062
- the maximum ratio of instance to spindles (see below for more
1063
  information) for local storage
1064

    
1065
All these together should allow all tools that talk to Ganeti to know
1066
what are the ranges of allowed values for instances and the
1067
over-subscription that is allowed.
1068

    
1069
For the VCPU/PCPU ratio, we already have the VCPU configuration from the
1070
instance configuration, and the physical CPU configuration from the
1071
node. For the spindle ratios however, we didn't track before these
1072
values, so new parameters have been added:
1073

    
1074
- a new node parameter ``spindle_count``, defaults to 1, customisable at
1075
  node group or node level
1076
- at new backend parameter (for instances), ``spindle_use`` defaults to 1
1077

    
1078
Note that spindles in this context doesn't need to mean actual
1079
mechanical hard-drives; it's just a relative number for both the node
1080
I/O capacity and instance I/O consumption.
1081

    
1082
Instance migration behaviour
1083
++++++++++++++++++++++++++++
1084

    
1085
While live-migration is in general desirable over failover, it is
1086
possible that for some workloads it is actually worse, due to the
1087
variable time of the โ€œsuspendโ€ phase during live migration.
1088

    
1089
To allow the tools to work consistently over such instances (without
1090
having to hard-code instance names), a new backend parameter
1091
``always_failover`` has been added to control the migration/failover
1092
behaviour. When set to True, all migration requests for an instance will
1093
instead fall-back to failover.
1094

    
1095
Instance memory ballooning
1096
++++++++++++++++++++++++++
1097

    
1098
Initial support for memory ballooning has been added. The memory for an
1099
instance is no longer fixed (backend parameter ``memory``), but instead
1100
can vary between minimum and maximum values (backend parameters
1101
``minmem`` and ``maxmem``). Currently we only change an instance's
1102
memory when:
1103

    
1104
- live migrating or failing over and instance and the target node
1105
  doesn't have enough memory
1106
- user requests changing the memory via ``gnt-instance modify
1107
  --runtime-memory``
1108

    
1109
Instance CPU pinning
1110
++++++++++++++++++++
1111

    
1112
In order to control the use of specific CPUs by instance, support for
1113
controlling CPU pinning has been added for the Xen, HVM and LXC
1114
hypervisors. This is controlled by a new hypervisor parameter
1115
``cpu_mask``; details about possible values for this are in the
1116
:manpage:`gnt-instance(8)`. Note that use of the most specific (precise
1117
VCPU-to-CPU mapping) form will work well only when all nodes in your
1118
cluster have the same amount of CPUs.
1119

    
1120
Disk parameters
1121
+++++++++++++++
1122

    
1123
Another area in which Ganeti was not customisable were the parameters
1124
used for storage configuration, e.g. how many stripes to use for LVM,
1125
DRBD resync configuration, etc.
1126

    
1127
To improve this area, we've added disks parameters, which are
1128
customisable at cluster and node group level, and which allow to
1129
specify various parameters for disks (DRBD has the most parameters
1130
currently), for example:
1131

    
1132
- DRBD resync algorithm and parameters (e.g. speed)
1133
- the default VG for meta-data volumes for DRBD
1134
- number of stripes for LVM (plain disk template)
1135
- the RBD pool
1136

    
1137
These parameters can be modified via ``gnt-cluster modify -D โ€ฆ`` and
1138
``gnt-group modify -D โ€ฆ``, and are used at either instance creation (in
1139
case of LVM stripes, for example) or at disk โ€œactivationโ€ time
1140
(e.g. resync speed).
1141

    
1142
Rados block device support
1143
++++++++++++++++++++++++++
1144

    
1145
A Rados (http://ceph.com/wiki/Rbd) storage backend has been added,
1146
denoted by the ``rbd`` disk template type. This is considered
1147
experimental, feedback is welcome. For details on configuring it, see
1148
the :doc:`install` document and the :manpage:`gnt-cluster(8)` man page.
1149

    
1150
Master IP setup
1151
+++++++++++++++
1152

    
1153
The existing master IP functionality works well only in simple setups (a
1154
single network shared by all nodes); however, if nodes belong to
1155
different networks, then the ``/32`` setup and lack of routing
1156
information is not enough.
1157

    
1158
To allow the master IP to function well in more complex cases, the
1159
system was reworked as follows:
1160

    
1161
- a master IP netmask setting has been added
1162
- the master IP activation/turn-down code was moved from the node daemon
1163
  to a separate script
1164
- whether to run the Ganeti-supplied master IP script or a user-supplied
1165
  on is a ``gnt-cluster init`` setting
1166

    
1167
Details about the location of the standard and custom setup scripts are
1168
in the man page :manpage:`gnt-cluster(8)`; for information about the
1169
setup script protocol, look at the Ganeti-supplied script.
1170

    
1171
SPICE support
1172
+++++++++++++
1173

    
1174
The `SPICE <http://www.linux-kvm.org/page/SPICE>`_ support has been
1175
improved.
1176

    
1177
It is now possible to use TLS-protected connections, and when renewing
1178
or changing the cluster certificates (via ``gnt-cluster renew-crypto``,
1179
it is now possible to specify spice or spice CA certificates. Also, it
1180
is possible to configure a password for SPICE sessions via the
1181
hypervisor parameter ``spice_password_file``.
1182

    
1183
There are also new parameters to control the compression and streaming
1184
options (e.g. ``spice_image_compression``, ``spice_streaming_video``,
1185
etc.). For details, see the man page :manpage:`gnt-instance(8)` and look
1186
for the spice parameters.
1187

    
1188
Lastly, it is now possible to see the SPICE connection information via
1189
``gnt-instance console``.
1190

    
1191
OVF converter
1192
+++++++++++++
1193

    
1194
A new tool (``tools/ovfconverter``) has been added that supports
1195
conversion between Ganeti and the `Open Virtualization Format
1196
<http://en.wikipedia.org/wiki/Open_Virtualization_Format>`_ (both to and
1197
from).
1198

    
1199
This relies on the ``qemu-img`` tool to convert the disk formats, so the
1200
actual compatibility with other virtualization solutions depends on it.
1201

    
1202
Confd daemon changes
1203
++++++++++++++++++++
1204

    
1205
The configuration query daemon (``ganeti-confd``) is now optional, and
1206
has been rewritten in Haskell; whether to use the daemon at all, use the
1207
Python (default) or the Haskell version is selectable at configure time
1208
via the ``--enable-confd`` parameter, which can take one of the
1209
``haskell``, ``python`` or ``no`` values. If not used, disabling the
1210
daemon will result in a smaller footprint; for larger systems, we
1211
welcome feedback on the Haskell version which might become the default
1212
in future versions.
1213

    
1214
If you want to use ``gnt-node list-drbd`` you need to have the Haskell
1215
daemon running. The Python version doesn't implement the new call.
1216

    
1217

    
1218
User interface changes
1219
~~~~~~~~~~~~~~~~~~~~~~
1220

    
1221
We have replaced the ``--disks`` option of ``gnt-instance
1222
replace-disks`` with a more flexible ``--disk`` option, which allows
1223
adding and removing disks at arbitrary indices (Issue 188). Furthermore,
1224
disk size and mode can be changed upon recreation (via ``gnt-instance
1225
recreate-disks``, which accepts the same ``--disk`` option).
1226

    
1227
As many people are used to a ``show`` command, we have added that as an
1228
alias to ``info`` on all ``gnt-*`` commands.
1229

    
1230
The ``gnt-instance grow-disk`` command has a new mode in which it can
1231
accept the target size of the disk, instead of the delta; this can be
1232
more safe since two runs in absolute mode will be idempotent, and
1233
sometimes it's also easier to specify the desired size directly.
1234

    
1235
Also the handling of instances with regard to offline secondaries has
1236
been improved. Instance operations should not fail because one of it's
1237
secondary nodes is offline, even though it's safe to proceed.
1238

    
1239
A new command ``list-drbd`` has been added to the ``gnt-node`` script to
1240
support debugging of DRBD issues on nodes. It provides a mapping of DRBD
1241
minors to instance name.
1242

    
1243
API changes
1244
~~~~~~~~~~~
1245

    
1246
RAPI coverage has improved, with (for example) new resources for
1247
recreate-disks, node power-cycle, etc.
1248

    
1249
Compatibility
1250
~~~~~~~~~~~~~
1251

    
1252
There is partial support for ``xl`` in the Xen hypervisor; feedback is
1253
welcome.
1254

    
1255
Python 2.7 is better supported, and after Ganeti 2.6 we will investigate
1256
whether to still support Python 2.4 or move to Python 2.6 as minimum
1257
required version.
1258

    
1259
Support for Fedora has been slightly improved; the provided example
1260
init.d script should work better on it and the INSTALL file should
1261
document the needed dependencies.
1262

    
1263
Internal changes
1264
~~~~~~~~~~~~~~~~
1265

    
1266
The deprecated ``QueryLocks`` LUXI request has been removed. Use
1267
``Query(what=QR_LOCK, ...)`` instead.
1268

    
1269
The LUXI requests :pyeval:`luxi.REQ_QUERY_JOBS`,
1270
:pyeval:`luxi.REQ_QUERY_INSTANCES`, :pyeval:`luxi.REQ_QUERY_NODES`,
1271
:pyeval:`luxi.REQ_QUERY_GROUPS`, :pyeval:`luxi.REQ_QUERY_EXPORTS` and
1272
:pyeval:`luxi.REQ_QUERY_TAGS` are deprecated and will be removed in a
1273
future version. :pyeval:`luxi.REQ_QUERY` should be used instead.
1274

    
1275
RAPI client: ``CertificateError`` now derives from
1276
``GanetiApiError``. This should make it more easy to handle Ganeti
1277
errors.
1278

    
1279
Deprecation warnings due to PyCrypto/paramiko import in
1280
``tools/setup-ssh`` have been silenced, as usually they are safe; please
1281
make sure to run an up-to-date paramiko version, if you use this tool.
1282

    
1283
The QA scripts now depend on Python 2.5 or above (the main code base
1284
still works with Python 2.4).
1285

    
1286
The configuration file (``config.data``) is now written without
1287
indentation for performance reasons; if you want to edit it, it can be
1288
re-formatted via ``tools/fmtjson``.
1289

    
1290
A number of bugs has been fixed in the cluster merge tool.
1291

    
1292
``x509`` certification verification (used in import-export) has been
1293
changed to allow the same clock skew as permitted by the cluster
1294
verification. This will remove some rare but hard to diagnose errors in
1295
import-export.
1296

    
1297

    
1298
Version 2.6.0 rc4
1299
-----------------
1300

    
1301
*(Released Thu, 19 Jul 2012)*
1302

    
1303
Very few changes from rc4 to the final release, only bugfixes:
1304

    
1305
- integrated fixes from release 2.5.2 (fix general boot flag for KVM
1306
  instance, fix CDROM booting for KVM instances)
1307
- fixed node group modification of node parameters
1308
- fixed issue in LUClusterVerifyGroup with multi-group clusters
1309
- fixed generation of bash completion to ensure a stable ordering
1310
- fixed a few typos
1311

    
1312

    
1313
Version 2.6.0 rc3
1314
-----------------
1315

    
1316
*(Released Fri, 13 Jul 2012)*
1317

    
1318
Third release candidate for 2.6. The following changes were done from
1319
rc3 to rc4:
1320

    
1321
- Fixed ``UpgradeConfig`` w.r.t. to disk parameters on disk objects.
1322
- Fixed an inconsistency in the LUXI protocol with the provided
1323
  arguments (NOT backwards compatible)
1324
- Fixed a bug with node groups ipolicy where ``min`` was greater than
1325
  the cluster ``std`` value
1326
- Implemented a new ``gnt-node list-drbd`` call to list DRBD minors for
1327
  easier instance debugging on nodes (requires ``hconfd`` to work)
1328

    
1329

    
1330
Version 2.6.0 rc2
1331
-----------------
1332

    
1333
*(Released Tue, 03 Jul 2012)*
1334

    
1335
Second release candidate for 2.6. The following changes were done from
1336
rc2 to rc3:
1337

    
1338
- Fixed ``gnt-cluster verify`` regarding ``master-ip-script`` on non
1339
  master candidates
1340
- Fixed a RAPI regression on missing beparams/memory
1341
- Fixed redistribution of files on offline nodes
1342
- Added possibility to run activate-disks even though secondaries are
1343
  offline. With this change it relaxes also the strictness on some other
1344
  commands which use activate disks internally:
1345
  * ``gnt-instance start|reboot|rename|backup|export``
1346
- Made it possible to remove safely an instance if its secondaries are
1347
  offline
1348
- Made it possible to reinstall even though secondaries are offline
1349

    
1350

    
1351
Version 2.6.0 rc1
1352
-----------------
1353

    
1354
*(Released Mon, 25 Jun 2012)*
1355

    
1356
First release candidate for 2.6. The following changes were done from
1357
rc1 to rc2:
1358

    
1359
- Fixed bugs with disk parameters and ``rbd`` templates as well as
1360
  ``instance_os_add``
1361
- Made ``gnt-instance modify`` more consistent regarding new NIC/Disk
1362
  behaviour. It supports now the modify operation
1363
- ``hcheck`` implemented to analyze cluster health and possibility of
1364
  improving health by rebalance
1365
- ``hbal`` has been improved in dealing with split instances
1366

    
1367

    
1368
Version 2.6.0 beta2
1369
-------------------
1370

    
1371
*(Released Mon, 11 Jun 2012)*
1372

    
1373
Second beta release of 2.6. The following changes were done from beta2
1374
to rc1:
1375

    
1376
- Fixed ``daemon-util`` with non-root user models
1377
- Fixed creation of plain instances with ``--no-wait-for-sync``
1378
- Fix wrong iv_names when running ``cfgupgrade``
1379
- Export more information in RAPI group queries
1380
- Fixed bug when changing instance network interfaces
1381
- Extended burnin to do NIC changes
1382
- query: Added ``<``, ``>``, ``<=``, ``>=`` comparison operators
1383
- Changed default for DRBD barriers
1384
- Fixed DRBD error reporting for syncer rate
1385
- Verify the options on disk parameters
1386

    
1387
And of course various fixes to documentation and improved unittests and
1388
QA.
1389

    
1390

    
1391
Version 2.6.0 beta1
1392
-------------------
1393

    
1394
*(Released Wed, 23 May 2012)*
1395

    
1396
First beta release of 2.6. The following changes were done from beta1 to
1397
beta2:
1398

    
1399
- integrated patch for distributions without ``start-stop-daemon``
1400
- adapted example init.d script to work on Fedora
1401
- fixed log handling in Haskell daemons
1402
- adapted checks in the watcher for pycurl linked against libnss
1403
- add partial support for ``xl`` instead of ``xm`` for Xen
1404
- fixed a type issue in cluster verification
1405
- fixed ssconf handling in the Haskell code (was breaking confd in IPv6
1406
  clusters)
1407

    
1408
Plus integrated fixes from the 2.5 branch:
1409

    
1410
- fixed ``kvm-ifup`` to use ``/bin/bash``
1411
- fixed parallel build failures
1412
- KVM live migration when using a custom keymap
1413

    
1414

    
1415
Version 2.5.2
1416
-------------
1417

    
1418
*(Released Tue, 24 Jul 2012)*
1419

    
1420
A small bugfix release, with no new features:
1421

    
1422
- fixed bash-isms in kvm-ifup, for compatibility with systems which use a
1423
  different default shell (e.g. Debian, Ubuntu)
1424
- fixed KVM startup and live migration with a custom keymap (fixes Issue
1425
  243 and Debian bug #650664)
1426
- fixed compatibility with KVM versions that don't support multiple boot
1427
  devices (fixes Issue 230 and Debian bug #624256)
1428

    
1429
Additionally, a few fixes were done to the build system (fixed parallel
1430
build failures) and to the unittests (fixed race condition in test for
1431
FileID functions, and the default enable/disable mode for QA test is now
1432
customisable).
1433

    
1434

    
1435
Version 2.5.1
1436
-------------
1437

    
1438
*(Released Fri, 11 May 2012)*
1439

    
1440
A small bugfix release.
1441

    
1442
The main issues solved are on the topic of compatibility with newer LVM
1443
releases:
1444

    
1445
- fixed parsing of ``lv_attr`` field
1446
- adapted to new ``vgreduce --removemissing`` behaviour where sometimes
1447
  the ``--force`` flag is needed
1448

    
1449
Also on the topic of compatibility, ``tools/lvmstrap`` has been changed
1450
to accept kernel 3.x too (was hardcoded to 2.6.*).
1451

    
1452
A regression present in 2.5.0 that broke handling (in the gnt-* scripts)
1453
of hook results and that also made display of other errors suboptimal
1454
was fixed; the code behaves now like 2.4 and earlier.
1455

    
1456
Another change in 2.5, the cleanup of the OS scripts environment, is too
1457
aggressive: it removed even the ``PATH`` variable, which requires the OS
1458
scripts to *always* need to export it. Since this is a bit too strict,
1459
we now export a minimal PATH, the same that we export for hooks.
1460

    
1461
The fix for issue 201 (Preserve bridge MTU in KVM ifup script) was
1462
integrated into this release.
1463

    
1464
Finally, a few other miscellaneous changes were done (no new features,
1465
just small improvements):
1466

    
1467
- Fix ``gnt-group --help`` display
1468
- Fix hardcoded Xen kernel path
1469
- Fix grow-disk handling of invalid units
1470
- Update synopsis for ``gnt-cluster repair-disk-sizes``
1471
- Accept both PUT and POST in noded (makes future upgrade to 2.6 easier)
1472

    
1473

    
1474
Version 2.5.0
1475
-------------
1476

    
1477
*(Released Thu, 12 Apr 2012)*
1478

    
1479
Incompatible/important changes and bugfixes
1480
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1481

    
1482
- The default of the ``/2/instances/[instance_name]/rename`` RAPI
1483
  resource's ``ip_check`` parameter changed from ``True`` to ``False``
1484
  to match the underlying LUXI interface.
1485
- The ``/2/nodes/[node_name]/evacuate`` RAPI resource was changed to use
1486
  body parameters, see :doc:`RAPI documentation <rapi>`. The server does
1487
  not maintain backwards-compatibility as the underlying operation
1488
  changed in an incompatible way. The RAPI client can talk to old
1489
  servers, but it needs to be told so as the return value changed.
1490
- When creating file-based instances via RAPI, the ``file_driver``
1491
  parameter no longer defaults to ``loop`` and must be specified.
1492
- The deprecated ``bridge`` NIC parameter is no longer supported. Use
1493
  ``link`` instead.
1494
- Support for the undocumented and deprecated RAPI instance creation
1495
  request format version 0 has been dropped. Use version 1, supported
1496
  since Ganeti 2.1.3 and :doc:`documented <rapi>`, instead.
1497
- Pyparsing 1.4.6 or above is required, see :doc:`installation
1498
  documentation <install>`.
1499
- The "cluster-verify" hooks are now executed per group by the
1500
  ``OP_CLUSTER_VERIFY_GROUP`` opcode. This maintains the same behavior
1501
  if you just run ``gnt-cluster verify``, which generates one opcode per
1502
  group.
1503
- The environment as passed to the OS scripts is cleared, and thus no
1504
  environment variables defined in the node daemon's environment will be
1505
  inherited by the scripts.
1506
- The :doc:`iallocator <iallocator>` mode ``multi-evacuate`` has been
1507
  deprecated.
1508
- :doc:`New iallocator modes <design-multi-reloc>` have been added to
1509
  support operations involving multiple node groups.
1510
- Offline nodes are ignored when failing over an instance.
1511
- Support for KVM version 1.0, which changed the version reporting format
1512
  from 3 to 2 digits.
1513
- TCP/IP ports used by DRBD disks are returned to a pool upon instance
1514
  removal.
1515
- ``Makefile`` is now compatible with Automake 1.11.2
1516
- Includes all bugfixes made in the 2.4 series
1517

    
1518
New features
1519
~~~~~~~~~~~~
1520

    
1521
- The ganeti-htools project has been merged into the ganeti-core source
1522
  tree and will be built as part of Ganeti (see :doc:`install-quick`).
1523
- Implemented support for :doc:`shared storage <design-shared-storage>`.
1524
- Add support for disks larger than 2 TB in ``lvmstrap`` by supporting
1525
  GPT-style partition tables (requires `parted
1526
  <http://www.gnu.org/s/parted/>`_).
1527
- Added support for floppy drive and 2nd CD-ROM drive in KVM hypervisor.
1528
- Allowed adding tags on instance creation.
1529
- Export instance tags to hooks (``INSTANCE_TAGS``, see :doc:`hooks`)
1530
- Allow instances to be started in a paused state, enabling the user to
1531
  see the complete console output on boot using the console.
1532
- Added new hypervisor flag to control default reboot behaviour
1533
  (``reboot_behavior``).
1534
- Added support for KVM keymaps (hypervisor parameter ``keymap``).
1535
- Improved out-of-band management support:
1536

    
1537
  - Added ``gnt-node health`` command reporting the health status of
1538
    nodes.
1539
  - Added ``gnt-node power`` command to manage power status of nodes.
1540
  - Added command for emergency power-off (EPO), ``gnt-cluster epo``.
1541

    
1542
- Instance migration can fall back to failover if instance is not
1543
  running.
1544
- Filters can be used when listing nodes, instances, groups and locks;
1545
  see :manpage:`ganeti(7)` manpage.
1546
- Added post-execution status as variables to :doc:`hooks <hooks>`
1547
  environment.
1548
- Instance tags are exported/imported together with the instance.
1549
- When given an explicit job ID, ``gnt-job info`` will work for archived
1550
  jobs.
1551
- Jobs can define dependencies on other jobs (not yet supported via
1552
  RAPI or command line, but used by internal commands and usable via
1553
  LUXI).
1554

    
1555
  - Lock monitor (``gnt-debug locks``) shows jobs waiting for
1556
    dependencies.
1557

    
1558
- Instance failover is now available as a RAPI resource
1559
  (``/2/instances/[instance_name]/failover``).
1560
- ``gnt-instance info`` defaults to static information if primary node
1561
  is offline.
1562
- Opcodes have a new ``comment`` attribute.
1563
- Added basic SPICE support to KVM hypervisor.
1564
- ``tools/ganeti-listrunner`` allows passing of arguments to executable.
1565

    
1566
Node group improvements
1567
~~~~~~~~~~~~~~~~~~~~~~~
1568

    
1569
- ``gnt-cluster verify`` has been modified to check groups separately,
1570
  thereby improving performance.
1571
- Node group support has been added to ``gnt-cluster verify-disks``,
1572
  which now operates per node group.
1573
- Watcher has been changed to work better with node groups.
1574

    
1575
  - One process and state file per node group.
1576
  - Slow watcher in one group doesn't block other group's watcher.
1577

    
1578
- Added new command, ``gnt-group evacuate``, to move all instances in a
1579
  node group to other groups.
1580
- Added ``gnt-instance change-group`` to move an instance to another
1581
  node group.
1582
- ``gnt-cluster command`` and ``gnt-cluster copyfile`` now support
1583
  per-group operations.
1584
- Node groups can be tagged.
1585
- Some operations switch from an exclusive to a shared lock as soon as
1586
  possible.
1587
- Instance's primary and secondary nodes' groups are now available as
1588
  query fields (``pnode.group``, ``pnode.group.uuid``, ``snodes.group``
1589
  and ``snodes.group.uuid``).
1590

    
1591
Misc
1592
~~~~
1593

    
1594
- Numerous updates to documentation and manpages.
1595

    
1596
  - :doc:`RAPI <rapi>` documentation now has detailed parameter
1597
    descriptions.
1598
  - Some opcode/job results are now also documented, see :doc:`RAPI
1599
    <rapi>`.
1600

    
1601
- A lockset's internal lock is now also visible in lock monitor.
1602
- Log messages from job queue workers now contain information about the
1603
  opcode they're processing.
1604
- ``gnt-instance console`` no longer requires the instance lock.
1605
- A short delay when waiting for job changes reduces the number of LUXI
1606
  requests significantly.
1607
- DRBD metadata volumes are overwritten with zeros during disk creation.
1608
- Out-of-band commands no longer acquire the cluster lock in exclusive
1609
  mode.
1610
- ``devel/upload`` now uses correct permissions for directories.
1611

    
1612

    
1613
Version 2.5.0 rc6
1614
-----------------
1615

    
1616
*(Released Fri, 23 Mar 2012)*
1617

    
1618
This was the sixth release candidate of the 2.5 series.
1619

    
1620

    
1621
Version 2.5.0 rc5
1622
-----------------
1623

    
1624
*(Released Mon, 9 Jan 2012)*
1625

    
1626
This was the fifth release candidate of the 2.5 series.
1627

    
1628

    
1629
Version 2.5.0 rc4
1630
-----------------
1631

    
1632
*(Released Thu, 27 Oct 2011)*
1633

    
1634
This was the fourth release candidate of the 2.5 series.
1635

    
1636

    
1637
Version 2.5.0 rc3
1638
-----------------
1639

    
1640
*(Released Wed, 26 Oct 2011)*
1641

    
1642
This was the third release candidate of the 2.5 series.
1643

    
1644

    
1645
Version 2.5.0 rc2
1646
-----------------
1647

    
1648
*(Released Tue, 18 Oct 2011)*
1649

    
1650
This was the second release candidate of the 2.5 series.
1651

    
1652

    
1653
Version 2.5.0 rc1
1654
-----------------
1655

    
1656
*(Released Tue, 4 Oct 2011)*
1657

    
1658
This was the first release candidate of the 2.5 series.
1659

    
1660

    
1661
Version 2.5.0 beta3
1662
-------------------
1663

    
1664
*(Released Wed, 31 Aug 2011)*
1665

    
1666
This was the third beta release of the 2.5 series.
1667

    
1668

    
1669
Version 2.5.0 beta2
1670
-------------------
1671

    
1672
*(Released Mon, 22 Aug 2011)*
1673

    
1674
This was the second beta release of the 2.5 series.
1675

    
1676

    
1677
Version 2.5.0 beta1
1678
-------------------
1679

    
1680
*(Released Fri, 12 Aug 2011)*
1681

    
1682
This was the first beta release of the 2.5 series.
1683

    
1684

    
1685
Version 2.4.5
1686
-------------
1687

    
1688
*(Released Thu, 27 Oct 2011)*
1689

    
1690
- Fixed bug when parsing command line parameter values ending in
1691
  backslash
1692
- Fixed assertion error after unclean master shutdown
1693
- Disable HTTP client pool for RPC, significantly reducing memory usage
1694
  of master daemon
1695
- Fixed queue archive creation with wrong permissions
1696

    
1697

    
1698
Version 2.4.4
1699
-------------
1700

    
1701
*(Released Tue, 23 Aug 2011)*
1702

    
1703
Small bug-fixes:
1704

    
1705
- Fixed documentation for importing with ``--src-dir`` option
1706
- Fixed a bug in ``ensure-dirs`` with queue/archive permissions
1707
- Fixed a parsing issue with DRBD 8.3.11 in the Linux kernel
1708

    
1709

    
1710
Version 2.4.3
1711
-------------
1712

    
1713
*(Released Fri, 5 Aug 2011)*
1714

    
1715
Many bug-fixes and a few small features:
1716

    
1717
- Fixed argument order in ``ReserveLV`` and ``ReserveMAC`` which caused
1718
  issues when you tried to add an instance with two MAC addresses in one
1719
  request
1720
- KVM: fixed per-instance stored UID value
1721
- KVM: configure bridged NICs at migration start
1722
- KVM: Fix a bug where instance will not start with never KVM versions
1723
  (>= 0.14)
1724
- Added OS search path to ``gnt-cluster info``
1725
- Fixed an issue with ``file_storage_dir`` where you were forced to
1726
  provide an absolute path, but the documentation states it is a
1727
  relative path, the documentation was right
1728
- Added a new parameter to instance stop/start called ``--no-remember``
1729
  that will make the state change to not be remembered
1730
- Implemented ``no_remember`` at RAPI level
1731
- Improved the documentation
1732
- Node evacuation: don't call IAllocator if node is already empty
1733
- Fixed bug in DRBD8 replace disks on current nodes
1734
- Fixed bug in recreate-disks for DRBD instances
1735
- Moved assertion checking locks in ``gnt-instance replace-disks``
1736
  causing it to abort with not owning the right locks for some situation
1737
- Job queue: Fixed potential race condition when cancelling queued jobs
1738
- Fixed off-by-one bug in job serial generation
1739
- ``gnt-node volumes``: Fix instance names
1740
- Fixed aliases in bash completion
1741
- Fixed a bug in reopening log files after being sent a SIGHUP
1742
- Added a flag to burnin to allow specifying VCPU count
1743
- Bugfixes to non-root Ganeti configuration
1744

    
1745

    
1746
Version 2.4.2
1747
-------------
1748

    
1749
*(Released Thu, 12 May 2011)*
1750

    
1751
Many bug-fixes and a few new small features:
1752

    
1753
- Fixed a bug related to log opening failures
1754
- Fixed a bug in instance listing with orphan instances
1755
- Fixed a bug which prevented resetting the cluster-level node parameter
1756
  ``oob_program`` to the default
1757
- Many fixes related to the ``cluster-merge`` tool
1758
- Fixed a race condition in the lock monitor, which caused failures
1759
  during (at least) creation of many instances in parallel
1760
- Improved output for gnt-job info
1761
- Removed the quiet flag on some ssh calls which prevented debugging
1762
  failures
1763
- Improved the N+1 failure messages in cluster verify by actually
1764
  showing the memory values (needed and available)
1765
- Increased lock attempt timeouts so that when executing long operations
1766
  (e.g. DRBD replace-disks) other jobs do not enter 'blocking acquire'
1767
  too early and thus prevent the use of the 'fair' mechanism
1768
- Changed instance query data (``gnt-instance info``) to not acquire
1769
  locks unless needed, thus allowing its use on locked instance if only
1770
  static information is asked for
1771
- Improved behaviour with filesystems that do not support rename on an
1772
  opened file
1773
- Fixed the behaviour of ``prealloc_wipe_disks`` cluster parameter which
1774
  kept locks on all nodes during the wipe, which is unneeded
1775
- Fixed ``gnt-watcher`` handling of errors during hooks execution
1776
- Fixed bug in ``prealloc_wipe_disks`` with small disk sizes (less than
1777
  10GiB) which caused the wipe to fail right at the end in some cases
1778
- Fixed master IP activation when doing master failover with no-voting
1779
- Fixed bug in ``gnt-node add --readd`` which allowed the re-adding of
1780
  the master node itself
1781
- Fixed potential data-loss in under disk full conditions, where Ganeti
1782
  wouldn't check correctly the return code and would consider
1783
  partially-written files 'correct'
1784
- Fixed bug related to multiple VGs and DRBD disk replacing
1785
- Added new disk parameter ``metavg`` that allows placement of the meta
1786
  device for DRBD in a different volume group
1787
- Fixed error handling in the node daemon when the system libc doesn't
1788
  have major number 6 (i.e. if ``libc.so.6`` is not the actual libc)
1789
- Fixed lock release during replace-disks, which kept cluster-wide locks
1790
  when doing disk replaces with an iallocator script
1791
- Added check for missing bridges in cluster verify
1792
- Handle EPIPE errors while writing to the terminal better, so that
1793
  piping the output to e.g. ``less`` doesn't cause a backtrace
1794
- Fixed rare case where a ^C during Luxi calls could have been
1795
  interpreted as server errors, instead of simply terminating
1796
- Fixed a race condition in LUGroupAssignNodes (``gnt-group
1797
  assign-nodes``)
1798
- Added a few more parameters to the KVM hypervisor, allowing a second
1799
  CDROM, custom disk type for CDROMs and a floppy image
1800
- Removed redundant message in instance rename when the name is given
1801
  already as a FQDN
1802
- Added option to ``gnt-instance recreate-disks`` to allow creating the
1803
  disks on new nodes, allowing recreation when the original instance
1804
  nodes are completely gone
1805
- Added option when converting disk templates to DRBD to skip waiting
1806
  for the resync, in order to make the instance available sooner
1807
- Added two new variables to the OS scripts environment (containing the
1808
  instance's nodes)
1809
- Made the root_path and optional parameter for the xen-pvm hypervisor,
1810
  to allow use of ``pvgrub`` as bootloader
1811
- Changed the instance memory modifications to only check out-of-memory
1812
  conditions on memory increases, and turned the secondary node warnings
1813
  into errors (they can still be overridden via ``--force``)
1814
- Fixed the handling of a corner case when the Python installation gets
1815
  corrupted (e.g. a bad disk) while ganeti-noded is running and we try
1816
  to execute a command that doesn't exist
1817
- Fixed a bug in ``gnt-instance move`` (LUInstanceMove) when the primary
1818
  node of the instance returned failures during instance shutdown; this
1819
  adds the option ``--ignore-consistency`` to gnt-instance move
1820

    
1821
And as usual, various improvements to the error messages, documentation
1822
and man pages.
1823

    
1824

    
1825
Version 2.4.1
1826
-------------
1827

    
1828
*(Released Wed, 09 Mar 2011)*
1829

    
1830
Emergency bug-fix release. ``tools/cfgupgrade`` was broken and overwrote
1831
the RAPI users file if run twice (even with ``--dry-run``).
1832

    
1833
The release fixes that bug (nothing else changed).
1834

    
1835

    
1836
Version 2.4.0
1837
-------------
1838

    
1839
*(Released Mon, 07 Mar 2011)*
1840

    
1841
Final 2.4.0 release. Just a few small fixes:
1842

    
1843
- Fixed RAPI node evacuate
1844
- Fixed the kvm-ifup script
1845
- Fixed internal error handling for special job cases
1846
- Updated man page to specify the escaping feature for options
1847

    
1848

    
1849
Version 2.4.0 rc3
1850
-----------------
1851

    
1852
*(Released Mon, 28 Feb 2011)*
1853

    
1854
A critical fix for the ``prealloc_wipe_disks`` feature: it is possible
1855
that this feature wiped the disks of the wrong instance, leading to loss
1856
of data.
1857

    
1858
Other changes:
1859

    
1860
- Fixed title of query field containing instance name
1861
- Expanded the glossary in the documentation
1862
- Fixed one unittest (internal issue)
1863

    
1864

    
1865
Version 2.4.0 rc2
1866
-----------------
1867

    
1868
*(Released Mon, 21 Feb 2011)*
1869

    
1870
A number of bug fixes plus just a couple functionality changes.
1871

    
1872
On the user-visible side, the ``gnt-* list`` command output has changed
1873
with respect to "special" field states. The current rc1 style of display
1874
can be re-enabled by passing a new ``--verbose`` (``-v``) flag, but in
1875
the default output mode special fields states are displayed as follows:
1876

    
1877
- Offline resource: ``*``
1878
- Unavailable/not applicable: ``-``
1879
- Data missing (RPC failure): ``?``
1880
- Unknown field: ``??``
1881

    
1882
Another user-visible change is the addition of ``--force-join`` to
1883
``gnt-node add``.
1884

    
1885
As for bug fixes:
1886

    
1887
- ``tools/cluster-merge`` has seen many fixes and is now enabled again
1888
- Fixed regression in RAPI/instance reinstall where all parameters were
1889
  required (instead of optional)
1890
- Fixed ``gnt-cluster repair-disk-sizes``, was broken since Ganeti 2.2
1891
- Fixed iallocator usage (offline nodes were not considered offline)
1892
- Fixed ``gnt-node list`` with respect to non-vm_capable nodes
1893
- Fixed hypervisor and OS parameter validation with respect to
1894
  non-vm_capable nodes
1895
- Fixed ``gnt-cluster verify`` with respect to offline nodes (mostly
1896
  cosmetic)
1897
- Fixed ``tools/listrunner`` with respect to agent-based usage
1898

    
1899

    
1900
Version 2.4.0 rc1
1901
-----------------
1902

    
1903
*(Released Fri,  4 Feb 2011)*
1904

    
1905
Many changes and fixes since the beta1 release. While there were some
1906
internal changes, the code has been mostly stabilised for the RC
1907
release.
1908

    
1909
Note: the dumb allocator was removed in this release, as it was not kept
1910
up-to-date with the IAllocator protocol changes. It is recommended to
1911
use the ``hail`` command from the ganeti-htools package.
1912

    
1913
Note: the 2.4 and up versions of Ganeti are not compatible with the
1914
0.2.x branch of ganeti-htools. You need to upgrade to
1915
ganeti-htools-0.3.0 (or later).
1916

    
1917
Regressions fixed from 2.3
1918
~~~~~~~~~~~~~~~~~~~~~~~~~~
1919

    
1920
- Fixed the ``gnt-cluster verify-disks`` command
1921
- Made ``gnt-cluster verify-disks`` work in parallel (as opposed to
1922
  serially on nodes)
1923
- Fixed disk adoption breakage
1924
- Fixed wrong headers in instance listing for field aliases
1925

    
1926
Other bugs fixed
1927
~~~~~~~~~~~~~~~~
1928

    
1929
- Fixed corner case in KVM handling of NICs
1930
- Fixed many cases of wrong handling of non-vm_capable nodes
1931
- Fixed a bug where a missing instance symlink was not possible to
1932
  recreate with any ``gnt-*`` command (now ``gnt-instance
1933
  activate-disks`` does it)
1934
- Fixed the volume group name as reported by ``gnt-cluster
1935
  verify-disks``
1936
- Increased timeouts for the import-export code, hopefully leading to
1937
  fewer aborts due network or instance timeouts
1938
- Fixed bug in ``gnt-node list-storage``
1939
- Fixed bug where not all daemons were started on cluster
1940
  initialisation, but only at the first watcher run
1941
- Fixed many bugs in the OOB implementation
1942
- Fixed watcher behaviour in presence of instances with offline
1943
  secondaries
1944
- Fixed instance list output for instances running on the wrong node
1945
- a few fixes to the cluster-merge tool, but it still cannot merge
1946
  multi-node groups (currently it is not recommended to use this tool)
1947

    
1948

    
1949
Improvements
1950
~~~~~~~~~~~~
1951

    
1952
- Improved network configuration for the KVM hypervisor
1953
- Added e1000 as a supported NIC for Xen-HVM
1954
- Improved the lvmstrap tool to also be able to use partitions, as
1955
  opposed to full disks
1956
- Improved speed of disk wiping (the cluster parameter
1957
  ``prealloc_wipe_disks``, so that it has a low impact on the total time
1958
  of instance creations
1959
- Added documentation for the OS parameters
1960
- Changed ``gnt-instance deactivate-disks`` so that it can work if the
1961
  hypervisor is not responding
1962
- Added display of blacklisted and hidden OS information in
1963
  ``gnt-cluster info``
1964
- Extended ``gnt-cluster verify`` to also validate hypervisor, backend,
1965
  NIC and node parameters, which might create problems with currently
1966
  invalid (but undetected) configuration files, but prevents validation
1967
  failures when unrelated parameters are modified
1968
- Changed cluster initialisation to wait for the master daemon to become
1969
  available
1970
- Expanded the RAPI interface:
1971

    
1972
  - Added config redistribution resource
1973
  - Added activation/deactivation of instance disks
1974
  - Added export of console information
1975

    
1976
- Implemented log file reopening on SIGHUP, which allows using
1977
  logrotate(8) for the Ganeti log files
1978
- Added a basic OOB helper script as an example
1979

    
1980

    
1981
Version 2.4.0 beta1
1982
-------------------
1983

    
1984
*(Released Fri, 14 Jan 2011)*
1985

    
1986
User-visible
1987
~~~~~~~~~~~~
1988

    
1989
- Fixed timezone issues when formatting timestamps
1990
- Added support for node groups, available via ``gnt-group`` and other
1991
  commands
1992
- Added out-of-band framework and management, see :doc:`design
1993
  document <design-oob>`
1994
- Removed support for roman numbers from ``gnt-node list`` and
1995
  ``gnt-instance list``.
1996
- Allowed modification of master network interface via ``gnt-cluster
1997
  modify --master-netdev``
1998
- Accept offline secondaries while shutting down instance disks
1999
- Added ``blockdev_prefix`` parameter to Xen PVM and HVM hypervisors
2000
- Added support for multiple LVM volume groups
2001
- Avoid sorting nodes for ``gnt-node list`` if specific nodes are
2002
  requested
2003
- Added commands to list available fields:
2004

    
2005
  - ``gnt-node list-fields``
2006
  - ``gnt-group list-fields``
2007
  - ``gnt-instance list-fields``
2008

    
2009
- Updated documentation and man pages
2010

    
2011
Integration
2012
~~~~~~~~~~~
2013

    
2014
- Moved ``rapi_users`` file into separate directory, now named
2015
  ``.../ganeti/rapi/users``, ``cfgupgrade`` moves the file and creates a
2016
  symlink
2017
- Added new tool for running commands on many machines,
2018
  ``tools/ganeti-listrunner``
2019
- Implemented more verbose result in ``OpInstanceConsole`` opcode, also
2020
  improving the ``gnt-instance console`` output
2021
- Allowed customisation of disk index separator at ``configure`` time
2022
- Export node group allocation policy to :doc:`iallocator <iallocator>`
2023
- Added support for non-partitioned md disks in ``lvmstrap``
2024
- Added script to gracefully power off KVM instances
2025
- Split ``utils`` module into smaller parts
2026
- Changed query operations to return more detailed information, e.g.
2027
  whether an information is unavailable due to an offline node. To use
2028
  this new functionality, the LUXI call ``Query`` must be used. Field
2029
  information is now stored by the master daemon and can be retrieved
2030
  using ``QueryFields``. Instances, nodes and groups can also be queried
2031
  using the new opcodes ``OpQuery`` and ``OpQueryFields`` (not yet
2032
  exposed via RAPI). The following commands make use of this
2033
  infrastructure change:
2034

    
2035
  - ``gnt-group list``
2036
  - ``gnt-group list-fields``
2037
  - ``gnt-node list``
2038
  - ``gnt-node list-fields``
2039
  - ``gnt-instance list``
2040
  - ``gnt-instance list-fields``
2041
  - ``gnt-debug locks``
2042

    
2043
Remote API
2044
~~~~~~~~~~
2045

    
2046
- New RAPI resources (see :doc:`rapi`):
2047

    
2048
  - ``/2/modify``
2049
  - ``/2/groups``
2050
  - ``/2/groups/[group_name]``
2051
  - ``/2/groups/[group_name]/assign-nodes``
2052
  - ``/2/groups/[group_name]/modify``
2053
  - ``/2/groups/[group_name]/rename``
2054
  - ``/2/instances/[instance_name]/disk/[disk_index]/grow``
2055

    
2056
- RAPI changes:
2057

    
2058
  - Implemented ``no_install`` for instance creation
2059
  - Implemented OS parameters for instance reinstallation, allowing
2060
    use of special settings on reinstallation (e.g. for preserving data)
2061

    
2062
Misc
2063
~~~~
2064

    
2065
- Added IPv6 support in import/export
2066
- Pause DRBD synchronization while wiping disks on instance creation
2067
- Updated unittests and QA scripts
2068
- Improved network parameters passed to KVM
2069
- Converted man pages from docbook to reStructuredText
2070

    
2071

    
2072
Version 2.3.1
2073
-------------
2074

    
2075
*(Released Mon, 20 Dec 2010)*
2076

    
2077
Released version 2.3.1~rc1 without any changes.
2078

    
2079

    
2080
Version 2.3.1 rc1
2081
-----------------
2082

    
2083
*(Released Wed, 1 Dec 2010)*
2084

    
2085
- impexpd: Disable OpenSSL compression in socat if possible (backport
2086
  from master, commit e90739d625b, see :doc:`installation guide
2087
  <install-quick>` for details)
2088
- Changed unittest coverage report to exclude test scripts
2089
- Added script to check version format
2090

    
2091

    
2092
Version 2.3.0
2093
-------------
2094

    
2095
*(Released Wed, 1 Dec 2010)*
2096

    
2097
Released version 2.3.0~rc1 without any changes.
2098

    
2099

    
2100
Version 2.3.0 rc1
2101
-----------------
2102

    
2103
*(Released Fri, 19 Nov 2010)*
2104

    
2105
A number of bugfixes and documentation updates:
2106

    
2107
- Update ganeti-os-interface documentation
2108
- Fixed a bug related to duplicate MACs or similar items which should be
2109
  unique
2110
- Fix breakage in OS state modify
2111
- Reinstall instance: disallow offline secondaries (fixes bug related to
2112
  OS changing but reinstall failing)
2113
- plus all the other fixes between 2.2.1 and 2.2.2
2114

    
2115

    
2116
Version 2.3.0 rc0
2117
-----------------
2118

    
2119
*(Released Tue, 2 Nov 2010)*
2120

    
2121
- Fixed clearing of the default iallocator using ``gnt-cluster modify``
2122
- Fixed master failover race with watcher
2123
- Fixed a bug in ``gnt-node modify`` which could lead to an inconsistent
2124
  configuration
2125
- Accept previously stopped instance for export with instance removal
2126
- Simplify and extend the environment variables for instance OS scripts
2127
- Added new node flags, ``master_capable`` and ``vm_capable``
2128
- Added optional instance disk wiping prior during allocation. This is a
2129
  cluster-wide option and can be set/modified using
2130
  ``gnt-cluster {init,modify} --prealloc-wipe-disks``.
2131
- Added IPv6 support, see :doc:`design document <design-2.3>` and
2132
  :doc:`install-quick`
2133
- Added a new watcher option (``--ignore-pause``)
2134
- Added option to ignore offline node on instance start/stop
2135
  (``--ignore-offline``)
2136
- Allow overriding OS parameters with ``gnt-instance reinstall``
2137
- Added ability to change node's secondary IP address using ``gnt-node
2138
  modify``
2139
- Implemented privilege separation for all daemons except
2140
  ``ganeti-noded``, see ``configure`` options
2141
- Complain if an instance's disk is marked faulty in ``gnt-cluster
2142
  verify``
2143
- Implemented job priorities (see ``ganeti(7)`` manpage)
2144
- Ignore failures while shutting down instances during failover from
2145
  offline node
2146
- Exit daemon's bootstrap process only once daemon is ready
2147
- Export more information via ``LUInstanceQuery``/remote API
2148
- Improved documentation, QA and unittests
2149
- RAPI daemon now watches ``rapi_users`` all the time and doesn't need a
2150
  restart if the file was created or changed
2151
- Added LUXI protocol version sent with each request and response,
2152
  allowing detection of server/client mismatches
2153
- Moved the Python scripts among gnt-* and ganeti-* into modules
2154
- Moved all code related to setting up SSH to an external script,
2155
  ``setup-ssh``
2156
- Infrastructure changes for node group support in future versions
2157

    
2158

    
2159
Version 2.2.2
2160
-------------
2161

    
2162
*(Released Fri, 19 Nov 2010)*
2163

    
2164
A few small bugs fixed, and some improvements to the build system:
2165

    
2166
- Fix documentation regarding conversion to drbd
2167
- Fix validation of parameters in cluster modify (``gnt-cluster modify
2168
  -B``)
2169
- Fix error handling in node modify with multiple changes
2170
- Allow remote imports without checked names
2171

    
2172

    
2173
Version 2.2.1
2174
-------------
2175

    
2176
*(Released Tue, 19 Oct 2010)*
2177

    
2178
- Disable SSL session ID cache in RPC client
2179

    
2180

    
2181
Version 2.2.1 rc1
2182
-----------------
2183

    
2184
*(Released Thu, 14 Oct 2010)*
2185

    
2186
- Fix interaction between Curl/GnuTLS and the Python's HTTP server
2187
  (thanks Apollon Oikonomopoulos!), finally allowing the use of Curl
2188
  with GnuTLS
2189
- Fix problems with interaction between Curl and Python's HTTP server,
2190
  resulting in increased speed in many RPC calls
2191
- Improve our release script to prevent breakage with older aclocal and
2192
  Python 2.6
2193

    
2194

    
2195
Version 2.2.1 rc0
2196
-----------------
2197

    
2198
*(Released Thu, 7 Oct 2010)*
2199

    
2200
- Fixed issue 125, replace hardcoded "xenvg" in ``gnt-cluster`` with
2201
  value retrieved from master
2202
- Added support for blacklisted or hidden OS definitions
2203
- Added simple lock monitor (accessible via (``gnt-debug locks``)
2204
- Added support for -mem-path in KVM hypervisor abstraction layer
2205
- Allow overriding instance parameters in tool for inter-cluster
2206
  instance moves (``tools/move-instance``)
2207
- Improved opcode summaries (e.g. in ``gnt-job list``)
2208
- Improve consistency of OS listing by sorting it
2209
- Documentation updates
2210

    
2211

    
2212
Version 2.2.0.1
2213
---------------
2214

    
2215
*(Released Fri, 8 Oct 2010)*
2216

    
2217
- Rebuild with a newer autotools version, to fix python 2.6 compatibility
2218

    
2219

    
2220
Version 2.2.0
2221
-------------
2222

    
2223
*(Released Mon, 4 Oct 2010)*
2224

    
2225
- Fixed regression in ``gnt-instance rename``
2226

    
2227

    
2228
Version 2.2.0 rc2
2229
-----------------
2230

    
2231
*(Released Wed, 22 Sep 2010)*
2232

    
2233
- Fixed OS_VARIANT variable for OS scripts
2234
- Fixed cluster tag operations via RAPI
2235
- Made ``setup-ssh`` exit with non-zero code if an error occurred
2236
- Disabled RAPI CA checks in watcher
2237

    
2238

    
2239
Version 2.2.0 rc1
2240
-----------------
2241

    
2242
*(Released Mon, 23 Aug 2010)*
2243

    
2244
- Support DRBD versions of the format "a.b.c.d"
2245
- Updated manpages
2246
- Re-introduce support for usage from multiple threads in RAPI client
2247
- Instance renames and modify via RAPI
2248
- Work around race condition between processing and archival in job
2249
  queue
2250
- Mark opcodes following failed one as failed, too
2251
- Job field ``lock_status`` was removed due to difficulties making it
2252
  work with the changed job queue in Ganeti 2.2; a better way to monitor
2253
  locks is expected for a later 2.2.x release
2254
- Fixed dry-run behaviour with many commands
2255
- Support ``ssh-agent`` again when adding nodes
2256
- Many additional bugfixes
2257

    
2258

    
2259
Version 2.2.0 rc0
2260
-----------------
2261

    
2262
*(Released Fri, 30 Jul 2010)*
2263

    
2264
Important change: the internal RPC mechanism between Ganeti nodes has
2265
changed from using a home-grown http library (based on the Python base
2266
libraries) to use the PycURL library. This requires that PycURL is
2267
installed on nodes. Please note that on Debian/Ubuntu, PycURL is linked
2268
against GnuTLS by default. cURL's support for GnuTLS had known issues
2269
before cURL 7.21.0 and we recommend using the latest cURL release or
2270
linking against OpenSSL. Most other distributions already link PycURL
2271
and cURL against OpenSSL. The command::
2272

    
2273
  python -c 'import pycurl; print pycurl.version'
2274

    
2275
can be used to determine the libraries PycURL and cURL are linked
2276
against.
2277

    
2278
Other significant changes:
2279

    
2280
- Rewrote much of the internals of the job queue, in order to achieve
2281
  better parallelism; this decouples job query operations from the job
2282
  processing, and it should allow much nicer behaviour of the master
2283
  daemon under load, and it also has uncovered some long-standing bugs
2284
  related to the job serialisation (now fixed)
2285
- Added a default iallocator setting to the cluster parameters,
2286
  eliminating the need to always pass nodes or an iallocator for
2287
  operations that require selection of new node(s)
2288
- Added experimental support for the LXC virtualization method
2289
- Added support for OS parameters, which allows the installation of
2290
  instances to pass parameter to OS scripts in order to customise the
2291
  instance
2292
- Added a hypervisor parameter controlling the migration type (live or
2293
  non-live), since hypervisors have various levels of reliability; this
2294
  has renamed the 'live' parameter to 'mode'
2295
- Added a cluster parameter ``reserved_lvs`` that denotes reserved
2296
  logical volumes, meaning that cluster verify will ignore them and not
2297
  flag their presence as errors
2298
- The watcher will now reset the error count for failed instances after
2299
  8 hours, thus allowing self-healing if the problem that caused the
2300
  instances to be down/fail to start has cleared in the meantime
2301
- Added a cluster parameter ``drbd_usermode_helper`` that makes Ganeti
2302
  check for, and warn, if the drbd module parameter ``usermode_helper``
2303
  is not consistent with the cluster-wide setting; this is needed to
2304
  make diagnose easier of failed drbd creations
2305
- Started adding base IPv6 support, but this is not yet
2306
  enabled/available for use
2307
- Rename operations (cluster, instance) will now return the new name,
2308
  which is especially useful if a short name was passed in
2309
- Added support for instance migration in RAPI
2310
- Added a tool to pre-configure nodes for the SSH setup, before joining
2311
  them to the cluster; this will allow in the future a simplified model
2312
  for node joining (but not yet fully enabled in 2.2); this needs the
2313
  paramiko python library
2314
- Fixed handling of name-resolving errors
2315
- Fixed consistency of job results on the error path
2316
- Fixed master-failover race condition when executed multiple times in
2317
  sequence
2318
- Fixed many bugs related to the job queue (mostly introduced during the
2319
  2.2 development cycle, so not all are impacting 2.1)
2320
- Fixed instance migration with missing disk symlinks
2321
- Fixed handling of unknown jobs in ``gnt-job archive``
2322
- And many other small fixes/improvements
2323

    
2324
Internal changes:
2325

    
2326
- Enhanced both the unittest and the QA coverage
2327
- Switched the opcode validation to a generic model, and extended the
2328
  validation to all opcode parameters
2329
- Changed more parts of the code that write shell scripts to use the
2330
  same class for this
2331
- Switched the master daemon to use the asyncore library for the Luxi
2332
  server endpoint
2333

    
2334

    
2335
Version 2.2.0 beta0
2336
-------------------
2337

    
2338
*(Released Thu, 17 Jun 2010)*
2339

    
2340
- Added tool (``move-instance``) and infrastructure to move instances
2341
  between separate clusters (see :doc:`separate documentation
2342
  <move-instance>` and :doc:`design document <design-2.2>`)
2343
- Added per-request RPC timeout
2344
- RAPI now requires a Content-Type header for requests with a body (e.g.
2345
  ``PUT`` or ``POST``) which must be set to ``application/json`` (see
2346
  :rfc:`2616` (HTTP/1.1), section 7.2.1)
2347
- ``ganeti-watcher`` attempts to restart ``ganeti-rapi`` if RAPI is not
2348
  reachable
2349
- Implemented initial support for running Ganeti daemons as separate
2350
  users, see configure-time flags ``--with-user-prefix`` and
2351
  ``--with-group-prefix`` (only ``ganeti-rapi`` is supported at this
2352
  time)
2353
- Instances can be removed after export (``gnt-backup export
2354
  --remove-instance``)
2355
- Self-signed certificates generated by Ganeti now use a 2048 bit RSA
2356
  key (instead of 1024 bit)
2357
- Added new cluster configuration file for cluster domain secret
2358
- Import/export now use SSL instead of SSH
2359
- Added support for showing estimated time when exporting an instance,
2360
  see the ``ganeti-os-interface(7)`` manpage and look for
2361
  ``EXP_SIZE_FD``
2362

    
2363

    
2364
Version 2.1.8
2365
-------------
2366

    
2367
*(Released Tue, 16 Nov 2010)*
2368

    
2369
Some more bugfixes. Unless critical bugs occur, this will be the last
2370
2.1 release:
2371

    
2372
- Fix case of MAC special-values
2373
- Fix mac checker regex
2374
- backend: Fix typo causing "out of range" error
2375
- Add missing --units in gnt-instance list man page
2376

    
2377

    
2378
Version 2.1.7
2379
-------------
2380

    
2381
*(Released Tue, 24 Aug 2010)*
2382

    
2383
Bugfixes only:
2384
  - Don't ignore secondary node silently on non-mirrored disk templates
2385
    (issue 113)
2386
  - Fix --master-netdev arg name in gnt-cluster(8) (issue 114)
2387
  - Fix usb_mouse parameter breaking with vnc_console (issue 109)
2388
  - Properly document the usb_mouse parameter
2389
  - Fix path in ganeti-rapi(8) (issue 116)
2390
  - Adjust error message when the ganeti user's .ssh directory is
2391
    missing
2392
  - Add same-node-check when changing the disk template to drbd
2393

    
2394

    
2395
Version 2.1.6
2396
-------------
2397

    
2398
*(Released Fri, 16 Jul 2010)*
2399

    
2400
Bugfixes only:
2401
  - Add an option to only select some reboot types during qa/burnin.
2402
    (on some hypervisors consequent reboots are not supported)
2403
  - Fix infrequent race condition in master failover. Sometimes the old
2404
    master ip address would be still detected as up for a short time
2405
    after it was removed, causing failover to fail.
2406
  - Decrease mlockall warnings when the ctypes module is missing. On
2407
    Python 2.4 we support running even if no ctypes module is installed,
2408
    but we were too verbose about this issue.
2409
  - Fix building on old distributions, on which man doesn't have a
2410
    --warnings option.
2411
  - Fix RAPI not to ignore the MAC address on instance creation
2412
  - Implement the old instance creation format in the RAPI client.
2413

    
2414

    
2415
Version 2.1.5
2416
-------------
2417

    
2418
*(Released Thu, 01 Jul 2010)*
2419

    
2420
A small bugfix release:
2421
  - Fix disk adoption: broken by strict --disk option checking in 2.1.4
2422
  - Fix batch-create: broken in the whole 2.1 series due to a lookup on
2423
    a non-existing option
2424
  - Fix instance create: the --force-variant option was ignored
2425
  - Improve pylint 0.21 compatibility and warnings with Python 2.6
2426
  - Fix modify node storage with non-FQDN arguments
2427
  - Fix RAPI client to authenticate under Python 2.6 when used
2428
    for more than 5 requests needing authentication
2429
  - Fix gnt-instance modify -t (storage) giving a wrong error message
2430
    when converting a non-shutdown drbd instance to plain
2431

    
2432

    
2433
Version 2.1.4
2434
-------------
2435

    
2436
*(Released Fri, 18 Jun 2010)*
2437

    
2438
A small bugfix release:
2439

    
2440
  - Fix live migration of KVM instances started with older Ganeti
2441
    versions which had fewer hypervisor parameters
2442
  - Fix gnt-instance grow-disk on down instances
2443
  - Fix an error-reporting bug during instance migration
2444
  - Better checking of the ``--net`` and ``--disk`` values, to avoid
2445
    silently ignoring broken ones
2446
  - Fix an RPC error reporting bug affecting, for example, RAPI client
2447
    users
2448
  - Fix bug triggered by different API version os-es on different nodes
2449
  - Fix a bug in instance startup with custom hvparams: OS level
2450
    parameters would fail to be applied.
2451
  - Fix the RAPI client under Python 2.6 (but more work is needed to
2452
    make it work completely well with OpenSSL)
2453
  - Fix handling of errors when resolving names from DNS
2454

    
2455

    
2456
Version 2.1.3
2457
-------------
2458

    
2459
*(Released Thu, 3 Jun 2010)*
2460

    
2461
A medium sized development cycle. Some new features, and some
2462
fixes/small improvements/cleanups.
2463

    
2464
Significant features
2465
~~~~~~~~~~~~~~~~~~~~
2466

    
2467
The node deamon now tries to mlock itself into memory, unless the
2468
``--no-mlock`` flag is passed. It also doesn't fail if it can't write
2469
its logs, and falls back to console logging. This allows emergency
2470
features such as ``gnt-node powercycle`` to work even in the event of a
2471
broken node disk (tested offlining the disk hosting the node's
2472
filesystem and dropping its memory caches; don't try this at home)
2473

    
2474
KVM: add vhost-net acceleration support. It can be tested with a new
2475
enough version of the kernel and of qemu-kvm.
2476

    
2477
KVM: Add instance chrooting feature. If you use privilege dropping for
2478
your VMs you can also now force them to chroot to an empty directory,
2479
before starting the emulated guest.
2480

    
2481
KVM: Add maximum migration bandwith and maximum downtime tweaking
2482
support (requires a new-enough version of qemu-kvm).
2483

    
2484
Cluster verify will now warn if the master node doesn't have the master
2485
ip configured on it.
2486

    
2487
Add a new (incompatible) instance creation request format to RAPI which
2488
supports all parameters (previously only a subset was supported, and it
2489
wasn't possible to extend the old format to accomodate all the new
2490
features. The old format is still supported, and a client can check for
2491
this feature, before using it, by checking for its presence in the
2492
``features`` RAPI resource.
2493

    
2494
Now with ancient latin support. Try it passing the ``--roman`` option to
2495
``gnt-instance info``, ``gnt-cluster info`` or ``gnt-node list``
2496
(requires the python-roman module to be installed, in order to work).
2497

    
2498
Other changes
2499
~~~~~~~~~~~~~
2500

    
2501
As usual many internal code refactorings, documentation updates, and
2502
such. Among others:
2503

    
2504
  - Lots of improvements and cleanups to the experimental Remote API
2505
    (RAPI) client library.
2506
  - A new unit test suite for the core daemon libraries.
2507
  - A fix to creating missing directories makes sure the umask is not
2508
    applied anymore. This enforces the same directory permissions
2509
    everywhere.
2510
  - Better handling terminating daemons with ctrl+c (used when running
2511
    them in debugging mode).
2512
  - Fix a race condition in live migrating a KVM instance, when stat()
2513
    on the old proc status file returned EINVAL, which is an unexpected
2514
    value.
2515
  - Fixed manpage checking with newer man and utf-8 charachters. But now
2516
    you need the en_US.UTF-8 locale enabled to build Ganeti from git.
2517

    
2518

    
2519
Version 2.1.2.1
2520
---------------
2521

    
2522
*(Released Fri, 7 May 2010)*
2523

    
2524
Fix a bug which prevented untagged KVM instances from starting.
2525

    
2526

    
2527
Version 2.1.2
2528
-------------
2529

    
2530
*(Released Fri, 7 May 2010)*
2531

    
2532
Another release with a long development cycle, during which many
2533
different features were added.
2534

    
2535
Significant features
2536
~~~~~~~~~~~~~~~~~~~~
2537

    
2538
The KVM hypervisor now can run the individual instances as non-root, to
2539
reduce the impact of a VM being hijacked due to bugs in the
2540
hypervisor. It is possible to run all instances as a single (non-root)
2541
user, to manually specify a user for each instance, or to dynamically
2542
allocate a user out of a cluster-wide pool to each instance, with the
2543
guarantee that no two instances will run under the same user ID on any
2544
given node.
2545

    
2546
An experimental RAPI client library, that can be used standalone
2547
(without the other Ganeti libraries), is provided in the source tree as
2548
``lib/rapi/client.py``. Note this client might change its interface in
2549
the future, as we iterate on its capabilities.
2550

    
2551
A new command, ``gnt-cluster renew-crypto`` has been added to easily
2552
replace the cluster's certificates and crypto keys. This might help in
2553
case they have been compromised, or have simply expired.
2554

    
2555
A new disk option for instance creation has been added that allows one
2556
to "adopt" currently existing logical volumes, with data
2557
preservation. This should allow easier migration to Ganeti from
2558
unmanaged (or managed via other software) instances.
2559

    
2560
Another disk improvement is the possibility to convert between redundant
2561
(DRBD) and plain (LVM) disk configuration for an instance. This should
2562
allow better scalability (starting with one node and growing the
2563
cluster, or shrinking a two-node cluster to one node).
2564

    
2565
A new feature that could help with automated node failovers has been
2566
implemented: if a node sees itself as offline (by querying the master
2567
candidates), it will try to shutdown (hard) all instances and any active
2568
DRBD devices. This reduces the risk of duplicate instances if an
2569
external script automatically failovers the instances on such nodes. To
2570
enable this, the cluster parameter ``maintain_node_health`` should be
2571
enabled; in the future this option (per the name) will enable other
2572
automatic maintenance features.
2573

    
2574
Instance export/import now will reuse the original instance
2575
specifications for all parameters; that means exporting an instance,
2576
deleting it and the importing it back should give an almost identical
2577
instance. Note that the default import behaviour has changed from
2578
before, where it created only one NIC; now it recreates the original
2579
number of NICs.
2580

    
2581
Cluster verify has added a few new checks: SSL certificates validity,
2582
/etc/hosts consistency across the cluster, etc.
2583

    
2584
Other changes
2585
~~~~~~~~~~~~~
2586

    
2587
As usual, many internal changes were done, documentation fixes,
2588
etc. Among others:
2589

    
2590
- Fixed cluster initialization with disabled cluster storage (regression
2591
  introduced in 2.1.1)
2592
- File-based storage supports growing the disks
2593
- Fixed behaviour of node role changes
2594
- Fixed cluster verify for some corner cases, plus a general rewrite of
2595
  cluster verify to allow future extension with more checks
2596
- Fixed log spamming by watcher and node daemon (regression introduced
2597
  in 2.1.1)
2598
- Fixed possible validation issues when changing the list of enabled
2599
  hypervisors
2600
- Fixed cleanup of /etc/hosts during node removal
2601
- Fixed RAPI response for invalid methods
2602
- Fixed bug with hashed passwords in ``ganeti-rapi`` daemon
2603
- Multiple small improvements to the KVM hypervisor (VNC usage, booting
2604
  from ide disks, etc.)
2605
- Allow OS changes without re-installation (to record a changed OS
2606
  outside of Ganeti, or to allow OS renames)
2607
- Allow instance creation without OS installation (useful for example if
2608
  the OS will be installed manually, or restored from a backup not in
2609
  Ganeti format)
2610
- Implemented option to make cluster ``copyfile`` use the replication
2611
  network
2612
- Added list of enabled hypervisors to ssconf (possibly useful for
2613
  external scripts)
2614
- Added a new tool (``tools/cfgupgrade12``) that allows upgrading from
2615
  1.2 clusters
2616
- A partial form of node re-IP is possible via node readd, which now
2617
  allows changed node primary IP
2618
- Command line utilities now show an informational message if the job is
2619
  waiting for a lock
2620
- The logs of the master daemon now show the PID/UID/GID of the
2621
  connected client
2622

    
2623

    
2624
Version 2.1.1
2625
-------------
2626

    
2627
*(Released Fri, 12 Mar 2010)*
2628

    
2629
During the 2.1.0 long release candidate cycle, a lot of improvements and
2630
changes have accumulated with were released later as 2.1.1.
2631

    
2632
Major changes
2633
~~~~~~~~~~~~~
2634

    
2635
The node evacuate command (``gnt-node evacuate``) was significantly
2636
rewritten, and as such the IAllocator protocol was changed - a new
2637
request type has been added. This unfortunate change during a stable
2638
series is designed to improve performance of node evacuations; on
2639
clusters with more than about five nodes and which are well-balanced,
2640
evacuation should proceed in parallel for all instances of the node
2641
being evacuated. As such, any existing IAllocator scripts need to be
2642
updated, otherwise the above command will fail due to the unknown
2643
request. The provided "dumb" allocator has not been updated; but the
2644
ganeti-htools package supports the new protocol since version 0.2.4.
2645

    
2646
Another important change is increased validation of node and instance
2647
names. This might create problems in special cases, if invalid host
2648
names are being used.
2649

    
2650
Also, a new layer of hypervisor parameters has been added, that sits at
2651
OS level between the cluster defaults and the instance ones. This allows
2652
customisation of virtualization parameters depending on the installed
2653
OS. For example instances with OS 'X' may have a different KVM kernel
2654
(or any other parameter) than the cluster defaults. This is intended to
2655
help managing a multiple OSes on the same cluster, without manual
2656
modification of each instance's parameters.
2657

    
2658
A tool for merging clusters, ``cluster-merge``, has been added in the
2659
tools sub-directory.
2660

    
2661
Bug fixes
2662
~~~~~~~~~
2663

    
2664
- Improved the int/float conversions that should make the code more
2665
  robust in face of errors from the node daemons
2666
- Fixed the remove node code in case of internal configuration errors
2667
- Fixed the node daemon behaviour in face of inconsistent queue
2668
  directory (e.g. read-only file-system where we can't open the files
2669
  read-write, etc.)
2670
- Fixed the behaviour of gnt-node modify for master candidate demotion;
2671
  now it either aborts cleanly or, if given the new "auto_promote"
2672
  parameter, will automatically promote other nodes as needed
2673
- Fixed compatibility with (unreleased yet) Python 2.6.5 that would
2674
  completely prevent Ganeti from working
2675
- Fixed bug for instance export when not all disks were successfully
2676
  exported
2677
- Fixed behaviour of node add when the new node is slow in starting up
2678
  the node daemon
2679
- Fixed handling of signals in the LUXI client, which should improve
2680
  behaviour of command-line scripts
2681
- Added checks for invalid node/instance names in the configuration (now
2682
  flagged during cluster verify)
2683
- Fixed watcher behaviour for disk activation errors
2684
- Fixed two potentially endless loops in http library, which led to the
2685
  RAPI daemon hanging and consuming 100% CPU in some cases
2686
- Fixed bug in RAPI daemon related to hashed passwords
2687
- Fixed bug for unintended qemu-level bridging of multi-NIC KVM
2688
  instances
2689
- Enhanced compatibility with non-Debian OSes, but not using absolute
2690
  path in some commands and allowing customisation of the ssh
2691
  configuration directory
2692
- Fixed possible future issue with new Python versions by abiding to the
2693
  proper use of ``__slots__`` attribute on classes
2694
- Added checks that should prevent directory traversal attacks
2695
- Many documentation fixes based on feedback from users
2696

    
2697
New features
2698
~~~~~~~~~~~~
2699

    
2700
- Added an "early_release" more for instance replace disks and node
2701
  evacuate, where we release locks earlier and thus allow higher
2702
  parallelism within the cluster
2703
- Added watcher hooks, intended to allow the watcher to restart other
2704
  daemons (e.g. from the ganeti-nbma project), but they can be used of
2705
  course for any other purpose
2706
- Added a compile-time disable for DRBD barriers, to increase
2707
  performance if the administrator trusts the power supply or the
2708
  storage system to not lose writes
2709
- Added the option of using syslog for logging instead of, or in
2710
  addition to, Ganeti's own log files
2711
- Removed boot restriction for paravirtual NICs for KVM, recent versions
2712
  can indeed boot from a paravirtual NIC
2713
- Added a generic debug level for many operations; while this is not
2714
  used widely yet, it allows one to pass the debug value all the way to
2715
  the OS scripts
2716
- Enhanced the hooks environment for instance moves (failovers,
2717
  migrations) where the primary/secondary nodes changed during the
2718
  operation, by adding {NEW,OLD}_{PRIMARY,SECONDARY} vars
2719
- Enhanced data validations for many user-supplied values; one important
2720
  item is the restrictions imposed on instance and node names, which
2721
  might reject some (invalid) host names
2722
- Add a configure-time option to disable file-based storage, if it's not
2723
  needed; this allows greater security separation between the master
2724
  node and the other nodes from the point of view of the inter-node RPC
2725
  protocol
2726
- Added user notification in interactive tools if job is waiting in the
2727
  job queue or trying to acquire locks
2728
- Added log messages when a job is waiting for locks
2729
- Added filtering by node tags in instance operations which admit
2730
  multiple instances (start, stop, reboot, reinstall)
2731
- Added a new tool for cluster mergers, ``cluster-merge``
2732
- Parameters from command line which are of the form ``a=b,c=d`` can now
2733
  use backslash escapes to pass in values which contain commas,
2734
  e.g. ``a=b\\c,d=e`` where the 'a' parameter would get the value
2735
  ``b,c``
2736
- For KVM, the instance name is the first parameter passed to KVM, so
2737
  that it's more visible in the process list
2738

    
2739

    
2740
Version 2.1.0
2741
-------------
2742

    
2743
*(Released Tue, 2 Mar 2010)*
2744

    
2745
Ganeti 2.1 brings many improvements with it. Major changes:
2746

    
2747
- Added infrastructure to ease automated disk repairs
2748
- Added new daemon to export configuration data in a cheaper way than
2749
  using the remote API
2750
- Instance NICs can now be routed instead of being associated with a
2751
  networking bridge
2752
- Improved job locking logic to reduce impact of jobs acquiring multiple
2753
  locks waiting for other long-running jobs
2754

    
2755
In-depth implementation details can be found in the Ganeti 2.1 design
2756
document.
2757

    
2758
Details
2759
~~~~~~~
2760

    
2761
- Added chroot hypervisor
2762
- Added more options to xen-hvm hypervisor (``kernel_path`` and
2763
  ``device_model``)
2764
- Added more options to xen-pvm hypervisor (``use_bootloader``,
2765
  ``bootloader_path`` and ``bootloader_args``)
2766
- Added the ``use_localtime`` option for the xen-hvm and kvm
2767
  hypervisors, and the default value for this has changed to false (in
2768
  2.0 xen-hvm always enabled it)
2769
- Added luxi call to submit multiple jobs in one go
2770
- Added cluster initialization option to not modify ``/etc/hosts``
2771
  file on nodes
2772
- Added network interface parameters
2773
- Added dry run mode to some LUs
2774
- Added RAPI resources:
2775

    
2776
  - ``/2/instances/[instance_name]/info``
2777
  - ``/2/instances/[instance_name]/replace-disks``
2778
  - ``/2/nodes/[node_name]/evacuate``
2779
  - ``/2/nodes/[node_name]/migrate``
2780
  - ``/2/nodes/[node_name]/role``
2781
  - ``/2/nodes/[node_name]/storage``
2782
  - ``/2/nodes/[node_name]/storage/modify``
2783
  - ``/2/nodes/[node_name]/storage/repair``
2784

    
2785
- Added OpCodes to evacuate or migrate all instances on a node
2786
- Added new command to list storage elements on nodes (``gnt-node
2787
  list-storage``) and modify them (``gnt-node modify-storage``)
2788
- Added new ssconf files with master candidate IP address
2789
  (``ssconf_master_candidates_ips``), node primary IP address
2790
  (``ssconf_node_primary_ips``) and node secondary IP address
2791
  (``ssconf_node_secondary_ips``)
2792
- Added ``ganeti-confd`` and a client library to query the Ganeti
2793
  configuration via UDP
2794
- Added ability to run hooks after cluster initialization and before
2795
  cluster destruction
2796
- Added automatic mode for disk replace (``gnt-instance replace-disks
2797
  --auto``)
2798
- Added ``gnt-instance recreate-disks`` to re-create (empty) disks
2799
  after catastrophic data-loss
2800
- Added ``gnt-node repair-storage`` command to repair damaged LVM volume
2801
  groups
2802
- Added ``gnt-instance move`` command to move instances
2803
- Added ``gnt-cluster watcher`` command to control watcher
2804
- Added ``gnt-node powercycle`` command to powercycle nodes
2805
- Added new job status field ``lock_status``
2806
- Added parseable error codes to cluster verification (``gnt-cluster
2807
  verify --error-codes``) and made output less verbose (use
2808
  ``--verbose`` to restore previous behaviour)
2809
- Added UUIDs to the main config entities (cluster, nodes, instances)
2810
- Added support for OS variants
2811
- Added support for hashed passwords in the Ganeti remote API users file
2812
  (``rapi_users``)
2813
- Added option to specify maximum timeout on instance shutdown
2814
- Added ``--no-ssh-init`` option to ``gnt-cluster init``
2815
- Added new helper script to start and stop Ganeti daemons
2816
  (``daemon-util``), with the intent to reduce the work necessary to
2817
  adjust Ganeti for non-Debian distributions and to start/stop daemons
2818
  from one place
2819
- Added more unittests
2820
- Fixed critical bug in ganeti-masterd startup
2821
- Removed the configure-time ``kvm-migration-port`` parameter, this is
2822
  now customisable at the cluster level for both the KVM and Xen
2823
  hypervisors using the new ``migration_port`` parameter
2824
- Pass ``INSTANCE_REINSTALL`` variable to OS installation script when
2825
  reinstalling an instance
2826
- Allowed ``@`` in tag names
2827
- Migrated to Sphinx (http://sphinx.pocoo.org/) for documentation
2828
- Many documentation updates
2829
- Distribute hypervisor files on ``gnt-cluster redist-conf``
2830
- ``gnt-instance reinstall`` can now reinstall multiple instances
2831
- Updated many command line parameters
2832
- Introduced new OS API version 15
2833
- No longer support a default hypervisor
2834
- Treat virtual LVs as inexistent
2835
- Improved job locking logic to reduce lock contention
2836
- Match instance and node names case insensitively
2837
- Reimplemented bash completion script to be more complete
2838
- Improved burnin
2839

    
2840

    
2841
Version 2.0.6
2842
-------------
2843

    
2844
*(Released Thu, 4 Feb 2010)*
2845

    
2846
- Fix cleaner behaviour on nodes not in a cluster (Debian bug 568105)
2847
- Fix a string formatting bug
2848
- Improve safety of the code in some error paths
2849
- Improve data validation in the master of values returned from nodes
2850

    
2851

    
2852
Version 2.0.5
2853
-------------
2854

    
2855
*(Released Thu, 17 Dec 2009)*
2856

    
2857
- Fix security issue due to missing validation of iallocator names; this
2858
  allows local and remote execution of arbitrary executables
2859
- Fix failure of gnt-node list during instance removal
2860
- Ship the RAPI documentation in the archive
2861

    
2862

    
2863
Version 2.0.4
2864
-------------
2865

    
2866
*(Released Wed, 30 Sep 2009)*
2867

    
2868
- Fixed many wrong messages
2869
- Fixed a few bugs related to the locking library
2870
- Fixed MAC checking at instance creation time
2871
- Fixed a DRBD parsing bug related to gaps in /proc/drbd
2872
- Fixed a few issues related to signal handling in both daemons and
2873
  scripts
2874
- Fixed the example startup script provided
2875
- Fixed insserv dependencies in the example startup script (patch from
2876
  Debian)
2877
- Fixed handling of drained nodes in the iallocator framework
2878
- Fixed handling of KERNEL_PATH parameter for xen-hvm (Debian bug
2879
  #528618)
2880
- Fixed error related to invalid job IDs in job polling
2881
- Fixed job/opcode persistence on unclean master shutdown
2882
- Fixed handling of partial job processing after unclean master
2883
  shutdown
2884
- Fixed error reporting from LUs, previously all errors were converted
2885
  into execution errors
2886
- Fixed error reporting from burnin
2887
- Decreased significantly the memory usage of the job queue
2888
- Optimised slightly multi-job submission
2889
- Optimised slightly opcode loading
2890
- Backported the multi-job submit framework from the development
2891
  branch; multi-instance start and stop should be faster
2892
- Added script to clean archived jobs after 21 days; this will reduce
2893
  the size of the queue directory
2894
- Added some extra checks in disk size tracking
2895
- Added an example ethers hook script
2896
- Added a cluster parameter that prevents Ganeti from modifying of
2897
  /etc/hosts
2898
- Added more node information to RAPI responses
2899
- Added a ``gnt-job watch`` command that allows following the ouput of a
2900
  job
2901
- Added a bind-address option to ganeti-rapi
2902
- Added more checks to the configuration verify
2903
- Enhanced the burnin script such that some operations can be retried
2904
  automatically
2905
- Converted instance reinstall to multi-instance model
2906

    
2907

    
2908
Version 2.0.3
2909
-------------
2910

    
2911
*(Released Fri, 7 Aug 2009)*
2912

    
2913
- Added ``--ignore-size`` to the ``gnt-instance activate-disks`` command
2914
  to allow using the pre-2.0.2 behaviour in activation, if any existing
2915
  instances have mismatched disk sizes in the configuration
2916
- Added ``gnt-cluster repair-disk-sizes`` command to check and update
2917
  any configuration mismatches for disk sizes
2918
- Added ``gnt-master cluste-failover --no-voting`` to allow master
2919
  failover to work on two-node clusters
2920
- Fixed the ``--net`` option of ``gnt-backup import``, which was
2921
  unusable
2922
- Fixed detection of OS script errors in ``gnt-backup export``
2923
- Fixed exit code of ``gnt-backup export``
2924

    
2925

    
2926
Version 2.0.2
2927
-------------
2928

    
2929
*(Released Fri, 17 Jul 2009)*
2930

    
2931
- Added experimental support for stripped logical volumes; this should
2932
  enhance performance but comes with a higher complexity in the block
2933
  device handling; stripping is only enabled when passing
2934
  ``--with-lvm-stripecount=N`` to ``configure``, but codepaths are
2935
  affected even in the non-stripped mode
2936
- Improved resiliency against transient failures at the end of DRBD
2937
  resyncs, and in general of DRBD resync checks
2938
- Fixed a couple of issues with exports and snapshot errors
2939
- Fixed a couple of issues in instance listing
2940
- Added display of the disk size in ``gnt-instance info``
2941
- Fixed checking for valid OSes in instance creation
2942
- Fixed handling of the "vcpus" parameter in instance listing and in
2943
  general of invalid parameters
2944
- Fixed http server library, and thus RAPI, to handle invalid
2945
  username/password combinations correctly; this means that now they
2946
  report unauthorized for queries too, not only for modifications,
2947
  allowing earlier detect of configuration problems
2948
- Added a new "role" node list field, equivalent to the master/master
2949
  candidate/drained/offline flags combinations
2950
- Fixed cluster modify and changes of candidate pool size
2951
- Fixed cluster verify error messages for wrong files on regular nodes
2952
- Fixed a couple of issues with node demotion from master candidate role
2953
- Fixed node readd issues
2954
- Added non-interactive mode for ``ganeti-masterd --no-voting`` startup
2955
- Added a new ``--no-voting`` option for masterfailover to fix failover
2956
  on two-nodes clusters when the former master node is unreachable
2957
- Added instance reinstall over RAPI
2958

    
2959

    
2960
Version 2.0.1
2961
-------------
2962

    
2963
*(Released Tue, 16 Jun 2009)*
2964

    
2965
- added ``-H``/``-B`` startup parameters to ``gnt-instance``, which will
2966
  allow re-adding the start in single-user option (regression from 1.2)
2967
- the watcher writes the instance status to a file, to allow monitoring
2968
  to report the instance status (from the master) based on cached
2969
  results of the watcher's queries; while this can get stale if the
2970
  watcher is being locked due to other work on the cluster, this is
2971
  still an improvement
2972
- the watcher now also restarts the node daemon and the rapi daemon if
2973
  they died
2974
- fixed the watcher to handle full and drained queue cases
2975
- hooks export more instance data in the environment, which helps if
2976
  hook scripts need to take action based on the instance's properties
2977
  (no longer need to query back into ganeti)
2978
- instance failovers when the instance is stopped do not check for free
2979
  RAM, so that failing over a stopped instance is possible in low memory
2980
  situations
2981
- rapi uses queries for tags instead of jobs (for less job traffic), and
2982
  for cluster tags it won't talk to masterd at all but read them from
2983
  ssconf
2984
- a couple of error handling fixes in RAPI
2985
- drbd handling: improved the error handling of inconsistent disks after
2986
  resync to reduce the frequency of "there are some degraded disks for
2987
  this instance" messages
2988
- fixed a bug in live migration when DRBD doesn't want to reconnect (the
2989
  error handling path called a wrong function name)
2990

    
2991

    
2992
Version 2.0.0
2993
-------------
2994

    
2995
*(Released Wed, 27 May 2009)*
2996

    
2997
- no changes from rc5
2998

    
2999

    
3000
Version 2.0 rc5
3001
---------------
3002

    
3003
*(Released Wed, 20 May 2009)*
3004

    
3005
- fix a couple of bugs (validation, argument checks)
3006
- fix ``gnt-cluster getmaster`` on non-master nodes (regression)
3007
- some small improvements to RAPI and IAllocator
3008
- make watcher automatically start the master daemon if down
3009

    
3010

    
3011
Version 2.0 rc4
3012
---------------
3013

    
3014
*(Released Mon, 27 Apr 2009)*
3015

    
3016
- change the OS list to not require locks; this helps with big clusters
3017
- fix ``gnt-cluster verify`` and ``gnt-cluster verify-disks`` when the
3018
  volume group is broken
3019
- ``gnt-instance info``, without any arguments, doesn't run for all
3020
  instances anymore; either pass ``--all`` or pass the desired
3021
  instances; this helps against mistakes on big clusters where listing
3022
  the information for all instances takes a long time
3023
- miscellaneous doc and man pages fixes
3024

    
3025

    
3026
Version 2.0 rc3
3027
---------------
3028

    
3029
*(Released Wed, 8 Apr 2009)*
3030

    
3031
- Change the internal locking model of some ``gnt-node`` commands, in
3032
  order to reduce contention (and blocking of master daemon) when
3033
  batching many creation/reinstall jobs
3034
- Fixes to Xen soft reboot
3035
- No longer build documentation at build time, instead distribute it in
3036
  the archive, in order to reduce the need for the whole docbook/rst
3037
  toolchains
3038

    
3039

    
3040
Version 2.0 rc2
3041
---------------
3042

    
3043
*(Released Fri, 27 Mar 2009)*
3044

    
3045
- Now the cfgupgrade scripts works and can upgrade 1.2.7 clusters to 2.0
3046
- Fix watcher startup sequence, improves the behaviour of busy clusters
3047
- Some other fixes in ``gnt-cluster verify``, ``gnt-instance
3048
  replace-disks``, ``gnt-instance add``, ``gnt-cluster queue``, KVM VNC
3049
  bind address and other places
3050
- Some documentation fixes and updates
3051

    
3052

    
3053
Version 2.0 rc1
3054
---------------
3055

    
3056
*(Released Mon, 2 Mar 2009)*
3057

    
3058
- More documentation updates, now all docs should be more-or-less
3059
  up-to-date
3060
- A couple of small fixes (mixed hypervisor clusters, offline nodes,
3061
  etc.)
3062
- Added a customizable HV_KERNEL_ARGS hypervisor parameter (for Xen PVM
3063
  and KVM)
3064
- Fix an issue related to $libdir/run/ganeti and cluster creation
3065

    
3066

    
3067
Version 2.0 beta2
3068
-----------------
3069

    
3070
*(Released Thu, 19 Feb 2009)*
3071

    
3072
- Xen PVM and KVM have switched the default value for the instance root
3073
  disk to the first partition on the first drive, instead of the whole
3074
  drive; this means that the OS installation scripts must be changed
3075
  accordingly
3076
- Man pages have been updated
3077
- RAPI has been switched by default to HTTPS, and the exported functions
3078
  should all work correctly
3079
- RAPI v1 has been removed
3080
- Many improvements to the KVM hypervisor
3081
- Block device errors are now better reported
3082
- Many other bugfixes and small improvements
3083

    
3084

    
3085
Version 2.0 beta1
3086
-----------------
3087

    
3088
*(Released Mon, 26 Jan 2009)*
3089

    
3090
- Version 2 is a general rewrite of the code and therefore the
3091
  differences are too many to list, see the design document for 2.0 in
3092
  the ``doc/`` subdirectory for more details
3093
- In this beta version there is not yet a migration path from 1.2 (there
3094
  will be one in the final 2.0 release)
3095
- A few significant changes are:
3096

    
3097
  - all commands are executed by a daemon (``ganeti-masterd``) and the
3098
    various ``gnt-*`` commands are just front-ends to it
3099
  - all the commands are entered into, and executed from a job queue,
3100
    see the ``gnt-job(8)`` manpage
3101
  - the RAPI daemon supports read-write operations, secured by basic
3102
    HTTP authentication on top of HTTPS
3103
  - DRBD version 0.7 support has been removed, DRBD 8 is the only
3104
    supported version (when migrating from Ganeti 1.2 to 2.0, you need
3105
    to migrate to DRBD 8 first while still running Ganeti 1.2)
3106
  - DRBD devices are using statically allocated minor numbers, which
3107
    will be assigned to existing instances during the migration process
3108
  - there is support for both Xen PVM and Xen HVM instances running on
3109
    the same cluster
3110
  - KVM virtualization is supported too
3111
  - file-based storage has been implemented, which means that it is
3112
    possible to run the cluster without LVM and DRBD storage, for
3113
    example using a shared filesystem exported from shared storage (and
3114
    still have live migration)
3115

    
3116

    
3117
Version 1.2.7
3118
-------------
3119

    
3120
*(Released Tue, 13 Jan 2009)*
3121

    
3122
- Change the default reboot type in ``gnt-instance reboot`` to "hard"
3123
- Reuse the old instance mac address by default on instance import, if
3124
  the instance name is the same.
3125
- Handle situations in which the node info rpc returns incomplete
3126
  results (issue 46)
3127
- Add checks for tcp/udp ports collisions in ``gnt-cluster verify``
3128
- Improved version of batcher:
3129

    
3130
  - state file support
3131
  - instance mac address support
3132
  - support for HVM clusters/instances
3133

    
3134
- Add an option to show the number of cpu sockets and nodes in
3135
  ``gnt-node list``
3136
- Support OSes that handle more than one version of the OS api (but do
3137
  not change the current API in any other way)
3138
- Fix ``gnt-node migrate``
3139
- ``gnt-debug`` man page
3140
- Fixes various more typos and small issues
3141
- Increase disk resync maximum speed to 60MB/s (from 30MB/s)
3142

    
3143

    
3144
Version 1.2.6
3145
-------------
3146

    
3147
*(Released Wed, 24 Sep 2008)*
3148

    
3149
- new ``--hvm-nic-type`` and ``--hvm-disk-type`` flags to control the
3150
  type of disk exported to fully virtualized instances.
3151
- provide access to the serial console of HVM instances
3152
- instance auto_balance flag, set by default. If turned off it will
3153
  avoid warnings on cluster verify if there is not enough memory to fail
3154
  over an instance. in the future it will prevent automatically failing
3155
  it over when we will support that.
3156
- batcher tool for instance creation, see ``tools/README.batcher``
3157
- ``gnt-instance reinstall --select-os`` to interactively select a new
3158
  operating system when reinstalling an instance.
3159
- when changing the memory amount on instance modify a check has been
3160
  added that the instance will be able to start. also warnings are
3161
  emitted if the instance will not be able to fail over, if auto_balance
3162
  is true.
3163
- documentation fixes
3164
- sync fields between ``gnt-instance list/modify/add/import``
3165
- fix a race condition in drbd when the sync speed was set after giving
3166
  the device a remote peer.
3167

    
3168

    
3169
Version 1.2.5
3170
-------------
3171

    
3172
*(Released Tue, 22 Jul 2008)*
3173

    
3174
- note: the allowed size and number of tags per object were reduced
3175
- fix a bug in ``gnt-cluster verify`` with inconsistent volume groups
3176
- fixed twisted 8.x compatibility
3177
- fixed ``gnt-instance replace-disks`` with iallocator
3178
- add TCP keepalives on twisted connections to detect restarted nodes
3179
- disk increase support, see ``gnt-instance grow-disk``
3180
- implement bulk node/instance query for RAPI
3181
- add tags in node/instance listing (optional)
3182
- experimental migration (and live migration) support, read the man page
3183
  for ``gnt-instance migrate``
3184
- the ``ganeti-watcher`` logs are now timestamped, and the watcher also
3185
  has some small improvements in handling its state file
3186

    
3187

    
3188
Version 1.2.4
3189
-------------
3190

    
3191
*(Released Fri, 13 Jun 2008)*
3192

    
3193
- Experimental readonly, REST-based remote API implementation;
3194
  automatically started on master node, TCP port 5080, if enabled by
3195
  ``--enable-rapi`` parameter to configure script.
3196
- Instance allocator support. Add and import instance accept a
3197
  ``--iallocator`` parameter, and call that instance allocator to decide
3198
  which node to use for the instance. The iallocator document describes
3199
  what's expected from an allocator script.
3200
- ``gnt-cluster verify`` N+1 memory redundancy checks: Unless passed the
3201
  ``--no-nplus1-mem`` option ``gnt-cluster verify`` now checks that if a
3202
  node is lost there is still enough memory to fail over the instances
3203
  that reside on it.
3204
- ``gnt-cluster verify`` hooks: it is now possible to add post-hooks to
3205
  ``gnt-cluster verify``, to check for site-specific compliance. All the
3206
  hooks will run, and their output, if any, will be displayed. Any
3207
  failing hook will make the verification return an error value.
3208
- ``gnt-cluster verify`` now checks that its peers are reachable on the
3209
  primary and secondary interfaces
3210
- ``gnt-node add`` now supports the ``--readd`` option, to readd a node
3211
  that is still declared as part of the cluster and has failed.
3212
- ``gnt-* list`` commands now accept a new ``-o +field`` way of
3213
  specifying output fields, that just adds the chosen fields to the
3214
  default ones.
3215
- ``gnt-backup`` now has a new ``remove`` command to delete an existing
3216
  export from the filesystem.
3217
- New per-instance parameters hvm_acpi, hvm_pae and hvm_cdrom_image_path
3218
  have been added. Using them you can enable/disable acpi and pae
3219
  support, and specify a path for a cd image to be exported to the
3220
  instance. These parameters as the name suggest only work on HVM
3221
  clusters.
3222
- When upgrading an HVM cluster to Ganeti 1.2.4, the values for ACPI and
3223
  PAE support will be set to the previously hardcoded values, but the
3224
  (previously hardcoded) path to the CDROM ISO image will be unset and
3225
  if required, needs to be set manually with ``gnt-instance modify``
3226
  after the upgrade.
3227
- The address to which an instance's VNC console is bound is now
3228
  selectable per-instance, rather than being cluster wide. Of course
3229
  this only applies to instances controlled via VNC, so currently just
3230
  applies to HVM clusters.
3231

    
3232

    
3233
Version 1.2.3
3234
-------------
3235

    
3236
*(Released Mon, 18 Feb 2008)*
3237

    
3238
- more tweaks to the disk activation code (especially helpful for DRBD)
3239
- change the default ``gnt-instance list`` output format, now there is
3240
  one combined status field (see the manpage for the exact values this
3241
  field will have)
3242
- some more fixes for the mac export to hooks change
3243
- make Ganeti not break with DRBD 8.2.x (which changed the version
3244
  format in ``/proc/drbd``) (issue 24)
3245
- add an upgrade tool from "remote_raid1" disk template to "drbd" disk
3246
  template, allowing migration from DRBD0.7+MD to DRBD8
3247

    
3248

    
3249
Version 1.2.2
3250
-------------
3251

    
3252
*(Released Wed, 30 Jan 2008)*
3253

    
3254
- fix ``gnt-instance modify`` breakage introduced in 1.2.1 with the HVM
3255
  support (issue 23)
3256
- add command aliases infrastructure and a few aliases
3257
- allow listing of VCPUs in the ``gnt-instance list`` and improve the
3258
  man pages and the ``--help`` option of ``gnt-node
3259
  list``/``gnt-instance list``
3260
- fix ``gnt-backup list`` with down nodes (issue 21)
3261
- change the tools location (move from $pkgdatadir to $pkglibdir/tools)
3262
- fix the dist archive and add a check for including svn/git files in
3263
  the future
3264
- some developer-related changes: improve the burnin and the QA suite,
3265
  add an upload script for testing during development
3266

    
3267

    
3268
Version 1.2.1
3269
-------------
3270

    
3271
*(Released Wed, 16 Jan 2008)*
3272

    
3273
- experimental HVM support, read the install document, section
3274
  "Initializing the cluster"
3275
- allow for the PVM hypervisor per-instance kernel and initrd paths
3276
- add a new command ``gnt-cluster verify-disks`` which uses a new
3277
  algorithm to improve the reconnection of the DRBD pairs if the device
3278
  on the secondary node has gone away
3279
- make logical volume code auto-activate LVs at disk activation time
3280
- slightly improve the speed of activating disks
3281
- allow specification of the MAC address at instance creation time, and
3282
  changing it later via ``gnt-instance modify``
3283
- fix handling of external commands that generate lots of output on
3284
  stderr
3285
- update documentation with regard to minimum version of DRBD8 supported
3286

    
3287

    
3288
Version 1.2.0
3289
-------------
3290

    
3291
*(Released Tue, 4 Dec 2007)*
3292

    
3293
- Log the ``xm create`` output to the node daemon log on failure (to
3294
  help diagnosing the error)
3295
- In debug mode, log all external commands output if failed to the logs
3296
- Change parsing of lvm commands to ignore stderr
3297

    
3298

    
3299
Version 1.2 beta3
3300
-----------------
3301

    
3302
*(Released Wed, 28 Nov 2007)*
3303

    
3304
- Another round of updates to the DRBD 8 code to deal with more failures
3305
  in the replace secondary node operation
3306
- Some more logging of failures in disk operations (lvm, drbd)
3307
- A few documentation updates
3308
- QA updates
3309

    
3310

    
3311
Version 1.2 beta2
3312
-----------------
3313

    
3314
*(Released Tue, 13 Nov 2007)*
3315

    
3316
- Change configuration file format from Python's Pickle to JSON.
3317
  Upgrading is possible using the cfgupgrade utility.
3318
- Add support for DRBD 8.0 (new disk template ``drbd``) which allows for
3319
  faster replace disks and is more stable (DRBD 8 has many improvements
3320
  compared to DRBD 0.7)
3321
- Added command line tags support (see man pages for ``gnt-instance``,
3322
  ``gnt-node``, ``gnt-cluster``)
3323
- Added instance rename support
3324
- Added multi-instance startup/shutdown
3325
- Added cluster rename support
3326
- Added ``gnt-node evacuate`` to simplify some node operations
3327
- Added instance reboot operation that can speedup reboot as compared to
3328
  stop and start
3329
- Soften the requirement that hostnames are in FQDN format
3330
- The ``ganeti-watcher`` now activates drbd pairs after secondary node
3331
  reboots
3332
- Removed dependency on debian's patched fping that uses the
3333
  non-standard ``-S`` option
3334
- Now the OS definitions are searched for in multiple, configurable
3335
  paths (easier for distros to package)
3336
- Some changes to the hooks infrastructure (especially the new
3337
  post-configuration update hook)
3338
- Other small bugfixes
3339

    
3340
.. vim: set textwidth=72 syntax=rst :
3341
.. Local Variables:
3342
.. mode: rst
3343
.. fill-column: 72
3344
.. End: