1 Ganeti administrator's guide
2 ============================
4 Documents Ganeti version |version|
13 Ganeti is a virtualization cluster management software. You are expected
14 to be a system administrator familiar with your Linux distribution and
15 the Xen or KVM virtualization environments before using it.
17 The various components of Ganeti all have man pages and interactive
18 help. This manual though will help you getting familiar with the system
19 by explaining the most common operations, grouped by related use.
21 After a terminology glossary and a section on the prerequisites needed
22 to use this manual, the rest of this document is divided in sections
23 for the different targets that a command affects: instance, nodes, etc.
25 .. _terminology-label:
30 This section provides a small introduction to Ganeti terminology, which
31 might be useful when reading the rest of the document.
36 A set of machines (nodes) that cooperate to offer a coherent, highly
37 available virtualization service under a single administration domain.
42 A physical machine which is member of a cluster. Nodes are the basic
43 cluster infrastructure, and they don't need to be fault tolerant in
44 order to achieve high availability for instances.
46 Node can be added and removed (if they host no instances) at will from
47 the cluster. In a HA cluster and only with HA instances, the loss of any
48 single node will not cause disk data loss for any instance; of course,
49 a node crash will cause the crash of the its primary instances.
51 A node belonging to a cluster can be in one of the following roles at a
54 - *master* node, which is the node from which the cluster is controlled
55 - *master candidate* node, only nodes in this role have the full cluster
56 configuration and knowledge, and only master candidates can become the
58 - *regular* node, which is the state in which most nodes will be on
59 bigger clusters (>20 nodes)
60 - *drained* node, nodes in this state are functioning normally but the
61 cannot receive new instances; the intention is that nodes in this role
62 have some issue and they are being evacuated for hardware repairs
63 - *offline* node, in which there is a record in the cluster
64 configuration about the node, but the daemons on the master node will
65 not talk to this node; any instances declared as having an offline
66 node as either primary or secondary will be flagged as an error in the
67 cluster verify operation
69 Depending on the role, each node will run a set of daemons:
71 - the :command:`ganeti-noded` daemon, which control the manipulation of
72 this node's hardware resources; it runs on all nodes which are in a
74 - the :command:`ganeti-confd` daemon (Ganeti 2.1+) which runs on all
75 nodes, but is only functional on master candidate nodes
76 - the :command:`ganeti-rapi` daemon which runs on the master node and
77 offers an HTTP-based API for the cluster
78 - the :command:`ganeti-masterd` daemon which runs on the master node and
79 allows control of the cluster
81 Beside the node role, there are other node flags that influence its
84 - the *master_capable* flag denotes whether the node can ever become a
85 master candidate; setting this to 'no' means that auto-promotion will
86 never make this node a master candidate; this flag can be useful for a
87 remote node that only runs local instances, and having it become a
88 master is impractical due to networking or other constraints
89 - the *vm_capable* flag denotes whether the node can host instances or
90 not; for example, one might use a non-vm_capable node just as a master
91 candidate, for configuration backups; setting this flag to no
92 disallows placement of instances of this node, deactivates hypervisor
93 and related checks on it (e.g. bridge checks, LVM check, etc.), and
94 removes it from cluster capacity computations
100 A virtual machine which runs on a cluster. It can be a fault tolerant,
101 highly available entity.
103 An instance has various parameters, which are classified in three
104 categories: hypervisor related-parameters (called ``hvparams``), general
105 parameters (called ``beparams``) and per network-card parameters (called
106 ``nicparams``). All these parameters can be modified either at instance
107 level or via defaults at cluster level.
112 The are multiple options for the storage provided to an instance; while
113 the instance sees the same virtual drive in all cases, the node-level
114 configuration varies between them.
116 There are four disk templates you can choose from:
119 The instance has no disks. Only used for special purpose operating
120 systems or for testing.
123 The instance will use plain files as backend for its disks. No
124 redundancy is provided, and this is somewhat more difficult to
125 configure for high performance.
128 The instance will use LVM devices as backend for its disks. No
129 redundancy is provided.
132 .. note:: This is only valid for multi-node clusters using DRBD 8.0+
134 A mirror is set between the local node and a remote one, which must be
135 specified with the second value of the --node option. Use this option
136 to obtain a highly available instance that can be failed over to a
137 remote node should the primary one fail.
142 A framework for using external (user-provided) scripts to compute the
143 placement of instances on the cluster nodes. This eliminates the need to
144 manually specify nodes in instance add, instance moves, node evacuate,
147 In order for Ganeti to be able to use these scripts, they must be place
148 in the iallocator directory (usually ``lib/ganeti/iallocators`` under
149 the installation prefix, e.g. ``/usr/local``).
151 “Primary” and “secondary” concepts
152 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
154 An instance has a primary and depending on the disk configuration, might
155 also have a secondary node. The instance always runs on the primary node
156 and only uses its secondary node for disk replication.
158 Similarly, the term of primary and secondary instances when talking
159 about a node refers to the set of instances having the given node as
160 primary, respectively secondary.
165 Tags are short strings that can be attached to either to cluster itself,
166 or to nodes or instances. They are useful as a very simplistic
167 information store for helping with cluster administration, for example
168 by attaching owner information to each instance after it's created::
170 gnt-instance add … instance1
171 gnt-instance add-tags instance1 owner:user2
173 And then by listing each instance and its tags, this information could
174 be used for contacting the users of each instance.
179 While not directly visible by an end-user, it's useful to know that a
180 basic cluster operation (e.g. starting an instance) is represented
181 internall by Ganeti as an *OpCode* (abbreviation from operation
182 code). These OpCodes are executed as part of a *Job*. The OpCodes in a
183 single Job are processed serially by Ganeti, but different Jobs will be
184 processed (depending on resource availability) in parallel. They will
185 not be executed in the submission order, but depending on resource
186 availability, locks and (starting with Ganeti 2.3) priority. An earlier
187 job may have to wait for a lock while a newer job doesn't need any locks
188 and can be executed right away. Operations requiring a certain order
189 need to be submitted as a single job, or the client must submit one job
190 at a time and wait for it to finish before continuing.
192 For example, shutting down the entire cluster can be done by running the
193 command ``gnt-instance shutdown --all``, which will submit for each
194 instance a separate job containing the “shutdown instance” OpCode.
200 You need to have your Ganeti cluster installed and configured before you
201 try any of the commands in this document. Please follow the
202 :doc:`install` for instructions on how to do that.
210 The add operation might seem complex due to the many parameters it
211 accepts, but once you have understood the (few) required parameters and
212 the customisation capabilities you will see it is an easy operation.
214 The add operation requires at minimum five parameters:
216 - the OS for the instance
218 - the disk count and size
219 - the node specification or alternatively the iallocator to use
220 - and finally the instance name
222 The OS for the instance must be visible in the output of the command
223 ``gnt-os list`` and specifies which guest OS to install on the instance.
225 The disk template specifies what kind of storage to use as backend for
226 the (virtual) disks presented to the instance; note that for instances
227 with multiple virtual disks, they all must be of the same type.
229 The node(s) on which the instance will run can be given either manually,
230 via the ``-n`` option, or computed automatically by Ganeti, if you have
231 installed any iallocator script.
233 With the above parameters in mind, the command is::
236 -n TARGET_NODE:SECONDARY_NODE \
238 -t DISK_TEMPLATE -s DISK_SIZE \
241 The instance name must be resolvable (e.g. exist in DNS) and usually
242 points to an address in the same subnet as the cluster itself.
244 The above command has the minimum required options; other options you
245 can give include, among others:
247 - The memory size (``-B memory``)
249 - The number of virtual CPUs (``-B vcpus``)
251 - Arguments for the NICs of the instance; by default, a single-NIC
252 instance is created. The IP and/or bridge of the NIC can be changed
253 via ``--nic 0:ip=IP,bridge=BRIDGE``
255 See the manpage for gnt-instance for the detailed option list.
257 For example if you want to create an highly available instance, with a
258 single disk of 50GB and the default memory size, having primary node
259 ``node1`` and secondary node ``node3``, use the following command::
261 gnt-instance add -n node1:node3 -o debootstrap -t drbd \
264 There is a also a command for batch instance creation from a
265 specification file, see the ``batch-create`` operation in the
266 gnt-instance manual page.
268 Regular instance operations
269 +++++++++++++++++++++++++++
274 Removing an instance is even easier than creating one. This operation is
275 irreversible and destroys all the contents of your instance. Use with
278 gnt-instance remove INSTANCE_NAME
283 Instances are automatically started at instance creation time. To
284 manually start one which is currently stopped you can run::
286 gnt-instance startup INSTANCE_NAME
288 While the command to stop one is::
290 gnt-instance shutdown INSTANCE_NAME
292 .. warning:: Do not use the Xen or KVM commands directly to stop
293 instances. If you run for example ``xm shutdown`` or ``xm destroy``
294 on an instance Ganeti will automatically restart it (via the
295 :command:`ganeti-watcher` command which is launched via cron).
300 There are two ways to get information about instances: listing
301 instances, which does a tabular output containing a given set of fields
302 about each instance, and querying detailed information about a set of
305 The command to see all the instances configured and their status is::
309 The command can return a custom set of information when using the ``-o``
310 option (as always, check the manpage for a detailed specification). Each
311 instance will be represented on a line, thus making it easy to parse
312 this output via the usual shell utilities (grep, sed, etc.).
314 To get more detailed information about an instance, you can run::
316 gnt-instance info INSTANCE
318 which will give a multi-line block of information about the instance,
319 it's hardware resources (especially its disks and their redundancy
320 status), etc. This is harder to parse and is more expensive than the
321 list operation, but returns much more detailed information.
327 You can create a snapshot of an instance disk and its Ganeti
328 configuration, which then you can backup, or import into another
329 cluster. The way to export an instance is::
331 gnt-backup export -n TARGET_NODE INSTANCE_NAME
334 The target node can be any node in the cluster with enough space under
335 ``/srv/ganeti`` to hold the instance image. Use the ``--noshutdown``
336 option to snapshot an instance without rebooting it. Note that Ganeti
337 only keeps one snapshot for an instance - any previous snapshot of the
338 same instance existing cluster-wide under ``/srv/ganeti`` will be
339 removed by this operation: if you want to keep them, you need to move
340 them out of the Ganeti exports directory.
342 Importing an instance is similar to creating a new one, but additionally
343 one must specify the location of the snapshot. The command is::
345 gnt-backup import -n TARGET_NODE \
346 --src-node=NODE --src-dir=DIR INSTANCE_NAME
348 By default, parameters will be read from the export information, but you
349 can of course pass them in via the command line - most of the options
350 available for the command :command:`gnt-instance add` are supported here
353 Import of foreign instances
354 +++++++++++++++++++++++++++
356 There is a possibility to import a foreign instance whose disk data is
357 already stored as LVM volumes without going through copying it: the disk
360 For this, ensure that the original, non-managed instance is stopped,
361 then create a Ganeti instance in the usual way, except that instead of
362 passing the disk information you specify the current volumes::
364 gnt-instance add -t plain -n HOME_NODE ... \
365 --disk 0:adopt=lv_name[,vg=vg_name] INSTANCE_NAME
367 This will take over the given logical volumes, rename them to the Ganeti
368 standard (UUID-based), and without installing the OS on them start
369 directly the instance. If you configure the hypervisor similar to the
370 non-managed configuration that the instance had, the transition should
371 be seamless for the instance. For more than one disk, just pass another
372 disk parameter (e.g. ``--disk 1:adopt=...``).
377 .. note:: This section only applies to multi-node clusters
379 .. _instance-change-primary-label:
381 Changing the primary node
382 +++++++++++++++++++++++++
384 There are three ways to exchange an instance's primary and secondary
385 nodes; the right one to choose depends on how the instance has been
386 created and the status of its current primary node. See
387 :ref:`rest-redundancy-label` for information on changing the secondary
388 node. Note that it's only possible to change the primary node to the
389 secondary and vice-versa; a direct change of the primary node with a
390 third node, while keeping the current secondary is not possible in a
391 single step, only via multiple operations as detailed in
392 :ref:`instance-relocation-label`.
394 Failing over an instance
395 ~~~~~~~~~~~~~~~~~~~~~~~~
397 If an instance is built in highly available mode you can at any time
398 fail it over to its secondary node, even if the primary has somehow
399 failed and it's not up anymore. Doing it is really easy, on the master
400 node you can just run::
402 gnt-instance failover INSTANCE_NAME
404 That's it. After the command completes the secondary node is now the
405 primary, and vice-versa.
407 Live migrating an instance
408 ~~~~~~~~~~~~~~~~~~~~~~~~~~
410 If an instance is built in highly available mode, it currently runs and
411 both its nodes are running fine, you can at migrate it over to its
412 secondary node, without downtime. On the master node you need to run::
414 gnt-instance migrate INSTANCE_NAME
416 The current load on the instance and its memory size will influence how
417 long the migration will take. In any case, for both KVM and Xen
418 hypervisors, the migration will be transparent to the instance.
420 Moving an instance (offline)
421 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
423 If an instance has not been create as mirrored, then the only way to
424 change its primary node is to execute the move command::
426 gnt-instance move -n NEW_NODE INSTANCE
428 This has a few prerequisites:
430 - the instance must be stopped
431 - its current primary node must be on-line and healthy
432 - the disks of the instance must not have any errors
434 Since this operation actually copies the data from the old node to the
435 new node, expect it to take proportional to the size of the instance's
436 disks and the speed of both the nodes' I/O system and their networking.
441 Disk failures are a common cause of errors in any server
442 deployment. Ganeti offers protection from single-node failure if your
443 instances were created in HA mode, and it also offers ways to restore
444 redundancy after a failure.
446 Preparing for disk operations
447 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
449 It is important to note that for Ganeti to be able to do any disk
450 operation, the Linux machines on top of which Ganeti must be consistent;
451 for LVM, this means that the LVM commands must not return failures; it
452 is common that after a complete disk failure, any LVM command aborts
453 with an error similar to::
456 /dev/sdb1: read failed after 0 of 4096 at 0: Input/output error
457 /dev/sdb1: read failed after 0 of 4096 at 750153695232: Input/output
459 /dev/sdb1: read failed after 0 of 4096 at 0: Input/output error
460 Couldn't find device with uuid
461 't30jmN-4Rcf-Fr5e-CURS-pawt-z0jU-m1TgeJ'.
462 Couldn't find all physical volumes for volume group xenvg.
464 Before restoring an instance's disks to healthy status, it's needed to
465 fix the volume group used by Ganeti so that we can actually create and
466 manage the logical volumes. This is usually done in a multi-step
469 #. first, if the disk is completely gone and LVM commands exit with
470 “Couldn't find device with uuid…” then you need to run the command::
472 vgreduce --removemissing VOLUME_GROUP
474 #. after the above command, the LVM commands should be executing
475 normally (warnings are normal, but the commands will not fail
478 #. if the failed disk is still visible in the output of the ``pvs``
479 command, you need to deactivate it from allocations by running::
483 At this point, the volume group should be consistent and any bad
484 physical volumes should not longer be available for allocation.
486 Note that since version 2.1 Ganeti provides some commands to automate
487 these two operations, see :ref:`storage-units-label`.
489 .. _rest-redundancy-label:
491 Restoring redundancy for DRBD-based instances
492 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
494 A DRBD instance has two nodes, and the storage on one of them has
495 failed. Depending on which node (primary or secondary) has failed, you
496 have three options at hand:
498 - if the storage on the primary node has failed, you need to re-create
500 - if the storage on the secondary node has failed, you can either
501 re-create the disks on it or change the secondary and recreate
502 redundancy on the new secondary node
504 Of course, at any point it's possible to force re-creation of disks even
505 though everything is already fine.
507 For all three cases, the ``replace-disks`` operation can be used::
509 # re-create disks on the primary node
510 gnt-instance replace-disks -p INSTANCE_NAME
511 # re-create disks on the current secondary
512 gnt-instance replace-disks -s INSTANCE_NAME
513 # change the secondary node, via manual specification
514 gnt-instance replace-disks -n NODE INSTANCE_NAME
515 # change the secondary node, via an iallocator script
516 gnt-instance replace-disks -I SCRIPT INSTANCE_NAME
517 # since Ganeti 2.1: automatically fix the primary or secondary node
518 gnt-instance replace-disks -a INSTANCE_NAME
520 Since the process involves copying all data from the working node to the
521 target node, it will take a while, depending on the instance's disk
522 size, node I/O system and network speed. But it is (baring any network
523 interruption) completely transparent for the instance.
525 Re-creating disks for non-redundant instances
526 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
528 .. versionadded:: 2.1
530 For non-redundant instances, there isn't a copy (except backups) to
531 re-create the disks. But it's possible to at-least re-create empty
532 disks, after which a reinstall can be run, via the ``recreate-disks``
535 gnt-instance recreate-disks INSTANCE
537 Note that this will fail if the disks already exists.
539 Conversion of an instance's disk type
540 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
542 It is possible to convert between a non-redundant instance of type
543 ``plain`` (LVM storage) and redundant ``drbd`` via the ``gnt-instance
546 # start with a non-redundant instance
547 gnt-instance add -t plain ... INSTANCE
549 # later convert it to redundant
550 gnt-instance stop INSTANCE
551 gnt-instance modify -t drbd -n NEW_SECONDARY INSTANCE
552 gnt-instance start INSTANCE
554 # and convert it back
555 gnt-instance stop INSTANCE
556 gnt-instance modify -t plain INSTANCE
557 gnt-instance start INSTANCE
559 The conversion must be done while the instance is stopped, and
560 converting from plain to drbd template presents a small risk, especially
561 if the instance has multiple disks and/or if one node fails during the
562 conversion procedure). As such, it's recommended (as always) to make
563 sure that downtime for manual recovery is acceptable and that the
564 instance has up-to-date backups.
569 Accessing an instance's disks
570 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
572 From an instance's primary node you can have access to its disks. Never
573 ever mount the underlying logical volume manually on a fault tolerant
574 instance, or will break replication and your data will be
575 inconsistent. The correct way to access an instance's disks is to run
576 (on the master node, as usual) the command::
578 gnt-instance activate-disks INSTANCE
580 And then, *on the primary node of the instance*, access the device that
581 gets created. For example, you could mount the given disks, then edit
582 files on the filesystem, etc.
584 Note that with partitioned disks (as opposed to whole-disk filesystems),
585 you will need to use a tool like :manpage:`kpartx(8)`::
587 node1# gnt-instance activate-disks instance1
590 node3# kpartx -l /dev/…
591 node3# kpartx -a /dev/…
592 node3# mount /dev/mapper/… /mnt/
593 # edit files under mnt as desired
595 node3# kpartx -d /dev/…
599 After you've finished you can deactivate them with the deactivate-disks
600 command, which works in the same way::
602 gnt-instance deactivate-disks INSTANCE
604 Note that if any process started by you is still using the disks, the
605 above command will error out, and you **must** cleanup and ensure that
606 the above command runs successfully before you start the instance,
607 otherwise the instance will suffer corruption.
609 Accessing an instance's console
610 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
612 The command to access a running instance's console is::
614 gnt-instance console INSTANCE_NAME
616 Use the console normally and then type ``^]`` when done, to exit.
618 Other instance operations
619 +++++++++++++++++++++++++
624 There is a wrapper command for rebooting instances::
626 gnt-instance reboot instance2
628 By default, this does the equivalent of shutting down and then starting
629 the instance, but it accepts parameters to perform a soft-reboot (via
630 the hypervisor), a hard reboot (hypervisor shutdown and then startup) or
631 a full one (the default, which also de-configures and then configures
632 again the disks of the instance).
634 Instance OS definitions debugging
635 +++++++++++++++++++++++++++++++++
637 Should you have any problems with instance operating systems the command
638 to see a complete status for all your nodes is::
642 .. _instance-relocation-label:
647 While it is not possible to move an instance from nodes ``(A, B)`` to
648 nodes ``(C, D)`` in a single move, it is possible to do so in a few
651 # instance is located on A, B
652 node1# gnt-instance replace -n nodeC instance1
653 # instance has moved from (A, B) to (A, C)
654 # we now flip the primary/secondary nodes
655 node1# gnt-instance migrate instance1
656 # instance lives on (C, A)
657 # we can then change A to D via:
658 node1# gnt-instance replace -n nodeD instance1
660 Which brings it into the final configuration of ``(C, D)``. Note that we
661 needed to do two replace-disks operation (two copies of the instance
662 disks), because we needed to get rid of both the original nodes (A and
668 There are much fewer node operations available than for instances, but
669 they are equivalently important for maintaining a healthy cluster.
674 It is at any time possible to extend the cluster with one more node, by
675 using the node add operation::
677 gnt-node add NEW_NODE
679 If the cluster has a replication network defined, then you need to pass
680 the ``-s REPLICATION_IP`` parameter to this option.
682 A variation of this command can be used to re-configure a node if its
683 Ganeti configuration is broken, for example if it has been reinstalled
686 gnt-node add --readd EXISTING_NODE
688 This will reinitialise the node as if it's been newly added, but while
689 keeping its existing configuration in the cluster (primary/secondary IP,
690 etc.), in other words you won't need to use ``-s`` here.
692 Changing the node role
693 ++++++++++++++++++++++
695 A node can be in different roles, as explained in the
696 :ref:`terminology-label` section. Promoting a node to the master role is
697 special, while the other roles are handled all via a single command.
699 Failing over the master node
700 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
702 If you want to promote a different node to the master role (for whatever
703 reason), run on any other master-candidate node the command::
705 gnt-cluster master-failover
707 and the node you ran it on is now the new master. In case you try to run
708 this on a non master-candidate node, you will get an error telling you
709 which nodes are valid.
711 Changing between the other roles
712 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
714 The ``gnt-node modify`` command can be used to select a new role::
716 # change to master candidate
717 gnt-node modify -C yes NODE
718 # change to drained status
719 gnt-node modify -D yes NODE
720 # change to offline status
721 gnt-node modify -O yes NODE
722 # change to regular mode (reset all flags)
723 gnt-node modify -O no -D no -C no NODE
725 Note that the cluster requires that at any point in time, a certain
726 number of nodes are master candidates, so changing from master candidate
727 to other roles might fail. It is recommended to either force the
728 operation (via the ``--force`` option) or first change the number of
729 master candidates in the cluster - see :ref:`cluster-config-label`.
734 There are two steps of moving instances off a node:
736 - moving the primary instances (actually converting them into secondary
738 - moving the secondary instances (including any instances converted in
741 Primary instance conversion
742 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
744 For this step, you can use either individual instance move
745 commands (as seen in :ref:`instance-change-primary-label`) or the bulk
746 per-node versions; these are::
748 gnt-node migrate NODE
749 gnt-node evacuate NODE
751 Note that the instance “move” command doesn't currently have a node
754 Both these commands, or the equivalent per-instance command, will make
755 this node the secondary node for the respective instances, whereas their
756 current secondary node will become primary. Note that it is not possible
757 to change in one step the primary node to another node as primary, while
758 keeping the same secondary node.
760 Secondary instance evacuation
761 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
763 For the evacuation of secondary instances, a command called
764 :command:`gnt-node evacuate` is provided and its syntax is::
766 gnt-node evacuate -I IALLOCATOR_SCRIPT NODE
767 gnt-node evacuate -n DESTINATION_NODE NODE
769 The first version will compute the new secondary for each instance in
770 turn using the given iallocator script, whereas the second one will
771 simply move all instances to DESTINATION_NODE.
776 Once a node no longer has any instances (neither primary nor secondary),
777 it's easy to remove it from the cluster::
779 gnt-node remove NODE_NAME
781 This will deconfigure the node, stop the ganeti daemons on it and leave
782 it hopefully like before it joined to the cluster.
787 When using LVM (either standalone or with DRBD), it can become tedious
788 to debug and fix it in case of errors. Furthermore, even file-based
789 storage can become complicated to handle manually on many hosts. Ganeti
790 provides a couple of commands to help with automation.
795 This is a command specific to LVM handling. It allows listing the
796 logical volumes on a given node or on all nodes and their association to
797 instances via the ``volumes`` command::
799 node1# gnt-node volumes
800 Node PhysDev VG Name Size Instance
801 node1 /dev/sdb1 xenvg e61fbc97-….disk0 512M instance17
802 node1 /dev/sdb1 xenvg ebd1a7d1-….disk0 512M instance19
803 node2 /dev/sdb1 xenvg 0af08a3d-….disk0 512M instance20
804 node2 /dev/sdb1 xenvg cc012285-….disk0 512M instance16
805 node2 /dev/sdb1 xenvg f0fac192-….disk0 512M instance18
807 The above command maps each logical volume to a volume group and
808 underlying physical volume and (possibly) to an instance.
810 .. _storage-units-label:
812 Generalized storage handling
813 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
815 .. versionadded:: 2.1
817 Starting with Ganeti 2.1, a new storage framework has been implemented
818 that tries to abstract the handling of the storage type the cluster
821 First is listing the backend storage and their space situation::
823 node1# gnt-node list-storage
824 Node Name Size Used Free
825 node1 /dev/sda7 673.8G 0M 673.8G
826 node1 /dev/sdb1 698.6G 1.5G 697.1G
827 node2 /dev/sda7 673.8G 0M 673.8G
828 node2 /dev/sdb1 698.6G 1.0G 697.6G
830 The default is to list LVM physical volumes. It's also possible to list
831 the LVM volume groups::
833 node1# gnt-node list-storage -t lvm-vg
838 Next is repairing storage units, which is currently only implemented for
839 volume groups and does the equivalent of ``vgreduce --removemissing``::
841 node1# gnt-node repair-storage node2 lvm-vg xenvg
842 Sun Oct 25 22:21:45 2009 Repairing storage unit 'xenvg' on node2 ...
844 Last is the modification of volume properties, which is (again) only
845 implemented for LVM physical volumes and allows toggling the
846 ``allocatable`` value::
848 node1# gnt-node modify-storage --allocatable=no node2 lvm-pv /dev/sdb1
850 Use of the storage commands
851 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
853 All these commands are needed when recovering a node from a disk
856 - first, we need to recover from complete LVM failure (due to missing
857 disk), by running the ``repair-storage`` command
858 - second, we need to change allocation on any partially-broken disk
859 (i.e. LVM still sees it, but it has bad blocks) by running
861 - then we can evacuate the instances as needed
867 Beside the cluster initialisation command (which is detailed in the
868 :doc:`install` document) and the master failover command which is
869 explained under node handling, there are a couple of other cluster
870 operations available.
872 .. _cluster-config-label:
877 One of the few commands that can be run on any node (not only the
878 master) is the ``getmaster`` command::
880 node2# gnt-cluster getmaster
884 It is possible to query and change global cluster parameters via the
885 ``info`` and ``modify`` commands::
887 node1# gnt-cluster info
888 Cluster name: cluster.example.com
889 Cluster UUID: 07805e6f-f0af-4310-95f1-572862ee939c
890 Creation time: 2009-09-25 05:04:15
891 Modification time: 2009-10-18 22:11:47
892 Master node: node1.example.com
893 Architecture (this node): 64bit (x86_64)
896 Default hypervisor: xen-pvm
897 Enabled hypervisors: xen-pvm
898 Hypervisor parameters:
903 - candidate pool size: 10
905 Default instance parameters:
909 Default nic parameters:
914 There various parameters above can be changed via the ``modify``
917 - the hypervisor parameters can be changed via ``modify -H
918 xen-pvm:root_path=…``, and so on for other hypervisors/key/values
919 - the "default instance parameters" are changeable via ``modify -B
920 parameter=value…`` syntax
921 - the cluster parameters are changeable via separate options to the
922 modify command (e.g. ``--candidate-pool-size``, etc.)
924 For detailed option list see the :manpage:`gnt-cluster(8)` man page.
926 The cluster version can be obtained via the ``version`` command::
927 node1# gnt-cluster version
928 Software version: 2.1.0
929 Internode protocol: 20
930 Configuration format: 2010000
934 This is not very useful except when debugging Ganeti.
939 There are two commands provided for replicating files to all nodes of a
940 cluster and for running commands on all the nodes::
942 node1# gnt-cluster copyfile /path/to/file
943 node1# gnt-cluster command ls -l /path/to/file
945 These are simple wrappers over scp/ssh and more advanced usage can be
946 obtained using :manpage:`dsh(1)` and similar commands. But they are
947 useful to update an OS script from the master node, for example.
952 There are three commands that relate to global cluster checks. The first
953 one is ``verify`` which gives an overview on the cluster state,
954 highlighting any issues. In normal operation, this command should return
955 no ``ERROR`` messages::
957 node1# gnt-cluster verify
958 Sun Oct 25 23:08:58 2009 * Verifying global settings
959 Sun Oct 25 23:08:58 2009 * Gathering data (2 nodes)
960 Sun Oct 25 23:09:00 2009 * Verifying node status
961 Sun Oct 25 23:09:00 2009 * Verifying instance status
962 Sun Oct 25 23:09:00 2009 * Verifying orphan volumes
963 Sun Oct 25 23:09:00 2009 * Verifying remaining instances
964 Sun Oct 25 23:09:00 2009 * Verifying N+1 Memory redundancy
965 Sun Oct 25 23:09:00 2009 * Other Notes
966 Sun Oct 25 23:09:00 2009 - NOTICE: 5 non-redundant instance(s) found.
967 Sun Oct 25 23:09:00 2009 * Hooks Results
969 The second command is ``verify-disks``, which checks that the instance's
970 disks have the correct status based on the desired instance state
973 node1# gnt-cluster verify-disks
975 Note that this command will show no output when disks are healthy.
977 The last command is used to repair any discrepancies in Ganeti's
978 recorded disk size and the actual disk size (disk size information is
979 needed for proper activation and growth of DRBD-based disks)::
981 node1# gnt-cluster repair-disk-sizes
982 Sun Oct 25 23:13:16 2009 - INFO: Disk 0 of instance instance1 has mismatched size, correcting: recorded 512, actual 2048
983 Sun Oct 25 23:13:17 2009 - WARNING: Invalid result from node node4, ignoring node results
985 The above shows one instance having wrong disk size, and a node which
986 returned invalid data, and thus we ignored all primary instances of that
989 Configuration redistribution
990 ++++++++++++++++++++++++++++
992 If the verify command complains about file mismatches between the master
993 and other nodes, due to some node problems or if you manually modified
994 configuration files, you can force an push of the master configuration
995 to all other nodes via the ``redist-conf`` command::
997 node1# gnt-cluster redist-conf
1000 This command will be silent unless there are problems sending updates to
1007 It is possible to rename a cluster, or to change its IP address, via the
1008 ``rename`` command. If only the IP has changed, you need to pass the
1009 current name and Ganeti will realise its IP has changed::
1011 node1# gnt-cluster rename cluster.example.com
1012 This will rename the cluster to 'cluster.example.com'. If
1013 you are connected over the network to the cluster name, the operation
1014 is very dangerous as the IP address will be removed from the node and
1015 the change may not go through. Continue?
1017 Failure: prerequisites not met for this operation:
1018 Neither the name nor the IP address of the cluster has changed
1020 In the above output, neither value has changed since the cluster
1021 initialisation so the operation is not completed.
1026 The job queue execution in Ganeti 2.0 and higher can be inspected,
1027 suspended and resumed via the ``queue`` command::
1029 node1~# gnt-cluster queue info
1030 The drain flag is unset
1031 node1~# gnt-cluster queue drain
1032 node1~# gnt-instance stop instance1
1033 Failed to submit job for instance1: Job queue is drained, refusing job
1034 node1~# gnt-cluster queue info
1035 The drain flag is set
1036 node1~# gnt-cluster queue undrain
1038 This is most useful if you have an active cluster and you need to
1039 upgrade the Ganeti software, or simply restart the software on any node:
1041 #. suspend the queue via ``queue drain``
1042 #. wait until there are no more running jobs via ``gnt-job list``
1043 #. restart the master or another node, or upgrade the software
1044 #. resume the queue via ``queue undrain``
1046 .. note:: this command only stores a local flag file, and if you
1047 failover the master, it will not have effect on the new master.
1053 The :manpage:`ganeti-watcher` is a program, usually scheduled via
1054 ``cron``, that takes care of cluster maintenance operations (restarting
1055 downed instances, activating down DRBD disks, etc.). However, during
1056 maintenance and troubleshooting, this can get in your way; disabling it
1057 via commenting out the cron job is not so good as this can be
1058 forgotten. Thus there are some commands for automated control of the
1059 watcher: ``pause``, ``info`` and ``continue``::
1061 node1~# gnt-cluster watcher info
1062 The watcher is not paused.
1063 node1~# gnt-cluster watcher pause 1h
1064 The watcher is paused until Mon Oct 26 00:30:37 2009.
1065 node1~# gnt-cluster watcher info
1066 The watcher is paused until Mon Oct 26 00:30:37 2009.
1067 node1~# ganeti-watcher -d
1068 2009-10-25 23:30:47,984: pid=28867 ganeti-watcher:486 DEBUG Pause has been set, exiting
1069 node1~# gnt-cluster watcher continue
1070 The watcher is no longer paused.
1071 node1~# ganeti-watcher -d
1072 2009-10-25 23:31:04,789: pid=28976 ganeti-watcher:345 DEBUG Archived 0 jobs, left 0
1073 2009-10-25 23:31:05,884: pid=28976 ganeti-watcher:280 DEBUG Got data from cluster, writing instance status file
1074 2009-10-25 23:31:06,061: pid=28976 ganeti-watcher:150 DEBUG Data didn't change, just touching status file
1075 node1~# gnt-cluster watcher info
1076 The watcher is not paused.
1079 The exact details of the argument to the ``pause`` command are available
1082 .. note:: this command only stores a local flag file, and if you
1083 failover the master, it will not have effect on the new master.
1085 Node auto-maintenance
1086 +++++++++++++++++++++
1088 If the cluster parameter ``maintain_node_health`` is enabled (see the
1089 manpage for :command:`gnt-cluster`, the init and modify subcommands),
1090 then the following will happen automatically:
1092 - the watcher will shutdown any instances running on offline nodes
1093 - the watcher will deactivate any DRBD devices on offline nodes
1095 In the future, more actions are planned, so only enable this parameter
1096 if the nodes are completely dedicated to Ganeti; otherwise it might be
1097 possible to lose data due to auto-maintenance actions.
1099 Removing a cluster entirely
1100 +++++++++++++++++++++++++++
1102 The usual method to cleanup a cluster is to run ``gnt-cluster destroy``
1103 however if the Ganeti installation is broken in any way then this will
1106 It is possible in such a case to cleanup manually most if not all traces
1107 of a cluster installation by following these steps on all of the nodes:
1109 1. Shutdown all instances. This depends on the virtualisation method
1110 used (Xen, KVM, etc.):
1112 - Xen: run ``xm list`` and ``xm destroy`` on all the non-Domain-0
1114 - KVM: kill all the KVM processes
1115 - chroot: kill all processes under the chroot mountpoints
1117 2. If using DRBD, shutdown all DRBD minors (which should by at this time
1118 no-longer in use by instances); on each node, run ``drbdsetup
1119 /dev/drbdN down`` for each active DRBD minor.
1121 3. If using LVM, cleanup the Ganeti volume group; if only Ganeti created
1122 logical volumes (and you are not sharing the volume group with the
1123 OS, for example), then simply running ``lvremove -f xenvg`` (replace
1124 'xenvg' with your volume group name) should do the required cleanup.
1126 4. If using file-based storage, remove recursively all files and
1127 directories under your file-storage directory: ``rm -rf
1128 /srv/ganeti/file-storage/*`` replacing the path with the correct path
1131 5. Stop the ganeti daemons (``/etc/init.d/ganeti stop``) and kill any
1132 that remain alive (``pgrep ganeti`` and ``pkill ganeti``).
1134 6. Remove the ganeti state directory (``rm -rf /var/lib/ganeti/*``),
1135 replacing the path with the correct path for your installation.
1137 On the master node, remove the cluster from the master-netdev (usually
1138 ``xen-br0`` for bridged mode, otherwise ``eth0`` or similar), by running
1139 ``ip a del $clusterip/32 dev xen-br0`` (use the correct cluster ip and
1140 network device name).
1142 At this point, the machines are ready for a cluster creation; in case
1143 you want to remove Ganeti completely, you need to also undo some of the
1144 SSH changes and log directories:
1146 - ``rm -rf /var/log/ganeti /srv/ganeti`` (replace with the correct
1148 - remove from ``/root/.ssh`` the keys that Ganeti added (check the
1149 ``authorized_keys`` and ``id_dsa`` files)
1150 - regenerate the host's SSH keys (check the OpenSSH startup scripts)
1153 Otherwise, if you plan to re-create the cluster, you can just go ahead
1154 and rerun ``gnt-cluster init``.
1159 The tags handling (addition, removal, listing) is similar for all the
1160 objects that support it (instances, nodes, and the cluster).
1165 Note that the set of characters present in a tag and the maximum tag
1166 length are restricted. Currently the maximum length is 128 characters,
1167 there can be at most 4096 tags per object, and the set of characters is
1168 comprised by alphanumeric characters and additionally ``.+*/:@-``.
1173 Tags can be added via ``add-tags``::
1175 gnt-instance add-tags INSTANCE a b c
1176 gnt-node add-tags INSTANCE a b c
1177 gnt-cluster add-tags a b c
1180 The above commands add three tags to an instance, to a node and to the
1181 cluster. Note that the cluster command only takes tags as arguments,
1182 whereas the node and instance commands first required the node and
1185 Tags can also be added from a file, via the ``--from=FILENAME``
1186 argument. The file is expected to contain one tag per line.
1188 Tags can also be remove via a syntax very similar to the add one::
1190 gnt-instance remove-tags INSTANCE a b c
1194 gnt-instance list-tags
1196 gnt-cluster list-tags
1201 It is also possible to execute a global search on the all tags defined
1202 in the cluster configuration, via a cluster command::
1204 gnt-cluster search-tags REGEXP
1206 The parameter expected is a regular expression (see
1207 :manpage:`regex(7)`). This will return all tags that match the search,
1208 together with the object they are defined in (the names being show in a
1209 hierarchical kind of way)::
1211 node1# gnt-cluster search-tags o
1213 /instances/instance1 owner:bar
1219 The various jobs submitted by the instance/node/cluster commands can be
1220 examined, canceled and archived by various invocations of the
1221 ``gnt-job`` command.
1223 First is the job list command::
1226 17771 success INSTANCE_QUERY_DATA
1227 17773 success CLUSTER_VERIFY_DISKS
1228 17775 success CLUSTER_REPAIR_DISK_SIZES
1229 17776 error CLUSTER_RENAME(cluster.example.com)
1230 17780 success CLUSTER_REDIST_CONF
1231 17792 success INSTANCE_REBOOT(instance1.example.com)
1233 More detailed information about a job can be found via the ``info``
1236 node1# gnt-job info 17776
1239 Received: 2009-10-25 23:18:02.180569
1240 Processing start: 2009-10-25 23:18:02.200335 (delta 0.019766s)
1241 Processing end: 2009-10-25 23:18:02.279743 (delta 0.079408s)
1242 Total processing time: 0.099174 seconds
1246 Processing start: 2009-10-25 23:18:02.200335
1247 Processing end: 2009-10-25 23:18:02.252282
1249 name: cluster.example.com
1252 [Neither the name nor the IP address of the cluster has changed]
1255 During the execution of a job, it's possible to follow the output of a
1256 job, similar to the log that one get from the ``gnt-`` commands, via the
1259 node1# gnt-instance add --submit … instance1
1261 node1# gnt-job watch 17818
1262 Output from job 17818 follows
1263 -----------------------------
1264 Mon Oct 26 00:22:48 2009 - INFO: Selected nodes for instance instance1 via iallocator dumb: node1, node2
1265 Mon Oct 26 00:22:49 2009 * creating instance disks...
1266 Mon Oct 26 00:22:52 2009 adding instance instance1 to cluster config
1267 Mon Oct 26 00:22:52 2009 - INFO: Waiting for instance instance1 to sync disks.
1269 Mon Oct 26 00:23:03 2009 creating os for instance instance1 on node node1
1270 Mon Oct 26 00:23:03 2009 * running the instance OS create scripts...
1271 Mon Oct 26 00:23:13 2009 * starting instance...
1274 This is useful if you need to follow a job's progress from multiple
1277 A job that has not yet started to run can be canceled::
1279 node1# gnt-job cancel 17810
1281 But not one that has already started execution::
1283 node1# gnt-job cancel 17805
1284 Job 17805 is no longer waiting in the queue
1286 There are two queues for jobs: the *current* and the *archive*
1287 queue. Jobs are initially submitted to the current queue, and they stay
1288 in that queue until they have finished execution (either successfully or
1289 not). At that point, they can be moved into the archive queue, and the
1290 ganeti-watcher script will do this automatically after 6 hours. The
1291 ganeti-cleaner script will remove the jobs from the archive directory
1294 Note that only jobs in the current queue can be viewed via the list and
1295 info commands; Ganeti itself doesn't examine the archive directory. If
1296 you need to see an older job, either move the file manually in the
1297 top-level queue directory, or look at its contents (it's a
1298 JSON-formatted file).
1300 Special Ganeti deployments
1301 --------------------------
1303 Since Ganeti 2.4, it is possible to extend the Ganeti deployment with
1304 two custom scenarios: Ganeti inside Ganeti and multi-site model.
1306 Running Ganeti under Ganeti
1307 +++++++++++++++++++++++++++
1309 It is sometimes useful to be able to use a Ganeti instance as a Ganeti
1310 node (part of another cluster, usually). One example scenario is two
1311 small clusters, where we want to have an additional master candidate
1312 that holds the cluster configuration and can be used for helping with
1313 the master voting process.
1315 However, these Ganeti instance should not host instances themselves, and
1316 should not be considered in the normal capacity planning, evacuation
1317 strategies, etc. In order to accomplish this, mark these nodes as
1318 non-``vm_capable``::
1320 node1# gnt-node modify --vm-capable=no node3
1322 The vm_capable status can be listed as usual via ``gnt-node list``::
1324 node1# gnt-node list -oname,vm_capable
1330 When this flag is set, the cluster will not do any operations that
1331 relate to instances on such nodes, e.g. hypervisor operations,
1332 disk-related operations, etc. Basically they will just keep the ssconf
1333 files, and if master candidates the full configuration.
1338 If Ganeti is deployed in multi-site model, with each site being a node
1339 group (so that instances are not relocated across the WAN by mistake),
1340 it is conceivable that either the WAN latency is high or that some sites
1341 have a lower reliability than others. In this case, it doesn't make
1342 sense to replicate the job information across all sites (or even outside
1343 of a “central” node group), so it should be possible to restrict which
1344 nodes can become master candidates via the auto-promotion algorithm.
1346 Ganeti 2.4 introduces for this purpose a new ``master_capable`` flag,
1347 which (when unset) prevents nodes from being marked as master
1348 candidates, either manually or automatically.
1350 As usual, the node modify operation can change this flag::
1352 node1# gnt-node modify --auto-promote --master-capable=no node3
1353 Fri Jan 7 06:23:07 2011 - INFO: Demoting from master candidate
1354 Fri Jan 7 06:23:08 2011 - INFO: Promoted nodes to master candidate role: node4
1356 - master_capable -> False
1357 - master_candidate -> False
1359 And the node list operation will list this flag::
1361 node1# gnt-node list -oname,master_capable node1 node2 node3
1367 Note that marking a node both not ``vm_capable`` and not
1368 ``master_capable`` makes the node practically unusable from Ganeti's
1369 point of view. Hence these two flags should be used probably in
1370 contrast: some nodes will be only master candidates (master_capable but
1371 not vm_capable), and other nodes will only hold instances (vm_capable
1372 but not master_capable).
1378 Beside the usual ``gnt-`` and ``ganeti-`` commands which are provided
1379 and installed in ``$prefix/sbin`` at install time, there are a couple of
1380 other tools installed which are used seldom but can be helpful in some
1386 The ``lvmstrap`` tool, introduced in :ref:`configure-lvm-label` section,
1387 has two modes of operation:
1389 - ``diskinfo`` shows the discovered disks on the system and their status
1390 - ``create`` takes all not-in-use disks and creates a volume group out
1393 .. warning:: The ``create`` argument to this command causes data-loss!
1398 The ``cfgupgrade`` tools is used to upgrade between major (and minor)
1399 Ganeti versions. Point-releases are usually transparent for the admin.
1401 More information about the upgrade procedure is listed on the wiki at
1402 http://code.google.com/p/ganeti/wiki/UpgradeNotes.
1404 There is also a script designed to upgrade from Ganeti 1.2 to 2.0,
1405 called ``cfgupgrade12``.
1410 .. note:: This command is not actively maintained; make sure you backup
1411 your configuration before using it
1413 This can be used as an alternative to direct editing of the
1414 main configuration file if Ganeti has a bug and prevents you, for
1415 example, from removing an instance or a node from the configuration
1423 .. warning:: This command will erase existing instances if given as
1426 This tool is used to exercise either the hardware of machines or
1427 alternatively the Ganeti software. It is safe to run on an existing
1428 cluster **as long as you don't pass it existing instance names**.
1430 The command will, by default, execute a comprehensive set of operations
1431 against a list of instances, these being:
1434 - disk replacement (for redundant instances)
1435 - failover and migration (for redundant instances)
1436 - move (for non-redundant instances)
1438 - add disks, remove disk
1439 - add NICs, remove NICs
1440 - export and then import
1444 - and finally removal of the test instances
1446 Executing all these operations will test that the hardware performs
1447 well: the creation, disk replace, disk add and disk growth will exercise
1448 the storage and network; the migrate command will test the memory of the
1449 systems. Depending on the passed options, it can also test that the
1450 instance OS definitions are executing properly the rename, import and
1456 This tool takes the Ganeti configuration and outputs a "sanitized"
1457 version, by randomizing or clearing:
1459 - DRBD secrets and cluster public key (always)
1460 - host names (optional)
1462 - OS names (optional)
1463 - LV names (optional, only useful for very old clusters which still have
1464 instances whose LVs are based on the instance name)
1466 By default, all optional items are activated except the LV name
1467 randomization. When passing ``--no-randomization``, which disables the
1468 optional items (i.e. just the DRBD secrets and cluster public keys are
1469 randomized), the resulting file can be used as a safety copy of the
1470 cluster config - while not trivial, the layout of the cluster can be
1471 recreated from it and if the instance disks have not been lost it
1472 permits recovery from the loss of all master candidates.
1477 See :doc:`separate documentation for move-instance <move-instance>`.
1479 .. TODO: document cluster-merge tool
1482 Other Ganeti projects
1483 ---------------------
1485 There are two other Ganeti-related projects that can be useful in a
1486 Ganeti deployment. These can be downloaded from the project site
1487 (http://code.google.com/p/ganeti/) and the repositories are also on the
1488 project git site (http://git.ganeti.org).
1493 The ``ganeti-nbma`` software is designed to allow instances to live on a
1494 separate, virtual network from the nodes, and in an environment where
1495 nodes are not guaranteed to be able to reach each other via multicasting
1496 or broadcasting. For more information see the README in the source
1502 The ``ganeti-htools`` software consists of a set of tools:
1504 - ``hail``: an advanced iallocator script compared to Ganeti's builtin
1506 - ``hbal``: a tool for rebalancing the cluster, i.e. moving instances
1507 around in order to better use the resources on the nodes
1508 - ``hspace``: a tool for estimating the available capacity of a cluster,
1509 so that capacity planning can be done efficiently
1511 For more information and installation instructions, see the README file
1512 in the source archive.
1514 .. vim: set textwidth=72 :