1 gnt-instance(8) Ganeti | Version @GANETI_VERSION@
2 =================================================
7 gnt-instance - Ganeti instance administration
12 **gnt-instance** {command} [arguments...]
17 The **gnt-instance** command is used for instance administration in
23 Creation/removal/querying
24 ~~~~~~~~~~~~~~~~~~~~~~~~~
30 | {-t|--disk-template {diskless | file \| plain \| drbd}}
31 | {--disk=*N*: {size=*VAL* \| adopt=*LV*}[,vg=*VG*][,metavg=*VG*][,mode=*ro\|rw*]
32 | \| {-s|--os-size} *SIZE*}
33 | [--no-ip-check] [--no-name-check] [--no-start] [--no-install]
34 | [--net=*N* [:options...] \| --no-nics]
35 | [{-B|--backend-parameters} *BEPARAMS*]
36 | [{-H|--hypervisor-parameters} *HYPERVISOR* [: option=*value*... ]]
37 | [{-O|--os-parameters} *param*=*value*... ]
38 | [--file-storage-dir *dir\_path*] [--file-driver {loop \| blktap}]
39 | {{-n|--node} *node[:secondary-node]* \| {-I|--iallocator} *name*}
40 | {{-o|--os-type} *os-type*}
44 Creates a new instance on the specified host. The *instance* argument
45 must be in DNS, but depending on the bridge/routing setup, need not be
46 in the same network as the nodes in the cluster.
48 The ``disk`` option specifies the parameters for the disks of the
49 instance. The numbering of disks starts at zero, and at least one disk
50 needs to be passed. For each disk, either the size or the adoption
51 source needs to be given, and optionally the access mode (read-only or
52 the default of read-write) and the LVM volume group can also be
53 specified (via the ``vg`` key). For DRBD devices, a different VG can
54 be specified for the metadata device using the ``metavg`` key. The
55 size is interpreted (when no unit is given) in mebibytes. You can also
56 use one of the suffixes *m*, *g* or *t* to specify the exact the units
57 used; these suffixes map to mebibytes, gibibytes and tebibytes.
59 When using the ``adopt`` key in the disk definition, Ganeti will
60 reuse those volumes (instead of creating new ones) as the
61 instance's disks. Ganeti will rename these volumes to the standard
62 format, and (without installing the OS) will use them as-is for the
63 instance. This allows migrating instances from non-managed mode
64 (e.g. plain KVM with LVM) to being managed via Ganeti. Please note that
65 this works only for the \`plain' disk template (see below for
68 Alternatively, a single-disk instance can be created via the ``-s``
69 option which takes a single argument, the size of the disk. This is
70 similar to the Ganeti 1.2 version (but will only create one disk).
72 The minimum disk specification is therefore ``--disk 0:size=20G`` (or
73 ``-s 20G`` when using the ``-s`` option), and a three-disk instance
74 can be specified as ``--disk 0:size=20G --disk 1:size=4G --disk
77 The ``--no-ip-check`` skips the checks that are done to see if the
78 instance's IP is not already alive (i.e. reachable from the master
81 The ``--no-name-check`` skips the check for the instance name via
82 the resolver (e.g. in DNS or /etc/hosts, depending on your setup).
83 Since the name check is used to compute the IP address, if you pass
84 this option you must also pass the ``--no-ip-check`` option.
86 If you don't wat the instance to automatically start after
87 creation, this is possible via the ``--no-start`` option. This will
88 leave the instance down until a subsequent **gnt-instance start**
91 The NICs of the instances can be specified via the ``--net``
92 option. By default, one NIC is created for the instance, with a
93 random MAC, and set up according the the cluster level nic
94 parameters. Each NIC can take these parameters (all optional):
97 either a value or 'generate' to generate a new unique MAC
100 specifies the IP address assigned to the instance from the Ganeti
101 side (this is not necessarily what the instance will use, but what
102 the node expects the instance to use)
105 specifies the connection mode for this nic: routed or bridged.
108 in bridged mode specifies the bridge to attach this NIC to, in
109 routed mode it's intended to differentiate between different
110 routing tables/instance groups (but the meaning is dependent on
111 the network script, see gnt-cluster(8) for more details)
114 Of these "mode" and "link" are nic parameters, and inherit their
115 default at cluster level. Alternatively, if no network is desired for
116 the instance, you can prevent the default of one NIC with the
117 ``--no-nics`` option.
119 The ``-o (--os-type)`` option specifies the operating system to be
120 installed. The available operating systems can be listed with
121 **gnt-os list**. Passing ``--no-install`` will however skip the OS
122 installation, allowing a manual import if so desired. Note that the
123 no-installation mode will automatically disable the start-up of the
124 instance (without an OS, it most likely won't be able to start-up
127 The ``-B (--backend-parameters)`` option specifies the backend
128 parameters for the instance. If no such parameters are specified, the
129 values are inherited from the cluster. Possible parameters are:
132 the maximum memory size of the instance; as usual, suffixes can be
133 used to denote the unit, otherwise the value is taken in mebibytes
136 the minimum memory size of the instance; as usual, suffixes can be
137 used to denote the unit, otherwise the value is taken in mebibytes
140 the number of VCPUs to assign to the instance (if this value makes
141 sense for the hypervisor)
144 whether the instance is considered in the N+1 cluster checks
145 (enough redundancy in the cluster to survive a node failure)
147 Note that before 2.6 Ganeti had a ``memory`` parameter, which was the
148 only value of memory an instance could have. With the
149 ``maxmem``/``minmem`` change Ganeti guarantees that at least the minimum
150 memory is always available for an instance, but allows more memory to be
151 used (up to the maximum memory) should it be free.
153 The ``-H (--hypervisor-parameters)`` option specified the hypervisor
154 to use for the instance (must be one of the enabled hypervisors on the
155 cluster) and optionally custom parameters for this instance. If not
156 other options are used (i.e. the invocation is just -H *NAME*) the
157 instance will inherit the cluster options. The defaults below show the
158 cluster defaults at cluster creation time.
160 The possible hypervisor options are as follows:
163 Valid for the Xen HVM and KVM hypervisors.
165 A string value denoting the boot order. This has different meaning
166 for the Xen HVM hypervisor and for the KVM one.
168 For Xen HVM, The boot order is a string of letters listing the boot
169 devices, with valid device letters being:
183 The default is not to set an HVM boot order, which is interpreted
186 For KVM the boot order is either "floppy", "cdrom", "disk" or
187 "network". Please note that older versions of KVM couldn't
188 netboot from virtio interfaces. This has been fixed in more recent
189 versions and is confirmed to work at least with qemu-kvm 0.11.1.
192 Valid for the Xen HVM and PVM hypervisors.
194 Relevant to non-pvops guest kernels, in which the disk device names
195 are given by the host. Allows one to specify 'xvd', which helps run
196 Red Hat based installers, driven by anaconda.
199 Valid for the KVM hypervisor.
201 The path to a floppy disk image to attach to the instance. This
202 is useful to install Windows operating systems on Virt/IO disks
203 because you can specify here the floppy for the drivers at
207 Valid for the Xen HVM and KVM hypervisors.
209 The path to a CDROM image to attach to the instance.
212 Valid for the KVM hypervisor.
214 The path to a second CDROM image to attach to the instance.
215 **NOTE**: This image can't be used to boot the system. To do that
216 you have to use the 'cdrom\_image\_path' option.
219 Valid for the Xen HVM and KVM hypervisors.
221 This parameter determines the way the network cards are presented
222 to the instance. The possible options are:
224 - rtl8139 (default for Xen HVM) (HVM & KVM)
225 - ne2k\_isa (HVM & KVM)
226 - ne2k\_pci (HVM & KVM)
232 - paravirtual (default for KVM) (HVM & KVM)
235 Valid for the Xen HVM and KVM hypervisors.
237 This parameter determines the way the disks are presented to the
238 instance. The possible options are:
240 - ioemu [default] (HVM & KVM)
249 Valid for the KVM hypervisor.
251 This parameter determines the way the cdroms disks are presented
252 to the instance. The default behavior is to get the same value of
253 the eariler parameter (disk_type). The possible options are:
264 Valid for the Xen HVM and KVM hypervisors.
266 Specifies the address that the VNC listener for this instance
267 should bind to. Valid values are IPv4 addresses. Use the address
268 0.0.0.0 to bind to all available interfaces (this is the default)
269 or specify the address of one of the interfaces on the node to
270 restrict listening to that interface.
273 Valid for the KVM hypervisor.
275 A boolean option that controls whether the VNC connection is
279 Valid for the KVM hypervisor.
281 If ``vnc_tls`` is enabled, this options specifies the path to the
282 x509 certificate to use.
285 Valid for the KVM hypervisor.
288 Valid for the KVM hypervisor.
290 Specifies the address or interface on which the SPICE server will
291 listen. Valid values are:
293 - IPv4 addresses, including 0.0.0.0 and 127.0.0.1
294 - IPv6 addresses, including :: and ::1
295 - names of network interfaces
297 If a network interface is specified, the SPICE server will be bound
298 to one of the addresses of that interface.
301 Valid for the KVM hypervisor.
303 Specifies which version of the IP protocol should be used by the
306 It is mainly intended to be used for specifying what kind of IP
307 addresses should be used if a network interface with both IPv4 and
308 IPv6 addresses is specified via the ``spice_bind`` parameter. In
309 this case, if the ``spice_ip_version`` parameter is not used, the
310 default IP version of the cluster will be used.
312 spice\_password\_file
313 Valid for the KVM hypervisor.
315 Specifies a file containing the password that must be used when
316 connecting via the SPICE protocol. If the option is not specified,
317 passwordless connections are allowed.
319 spice\_image\_compression
320 Valid for the KVM hypervisor.
322 Configures the SPICE lossless image compression. Valid values are:
331 spice\_jpeg\_wan\_compression
332 Valid for the KVM hypervisor.
334 Configures how SPICE should use the jpeg algorithm for lossy image
335 compression on slow links. Valid values are:
341 spice\_zlib\_glz\_wan\_compression
342 Valid for the KVM hypervisor.
344 Configures how SPICE should use the zlib-glz algorithm for lossy image
345 compression on slow links. Valid values are:
351 spice\_streaming\_video
352 Valid for the KVM hypervisor.
354 Configures how SPICE should detect video streams. Valid values are:
360 spice\_playback\_compression
361 Valid for the KVM hypervisor.
363 Configures whether SPICE should compress audio streams or not.
366 Valid for the KVM hypervisor.
368 Specifies that the SPICE server must use TLS to encrypt all the
369 traffic with the client.
372 Valid for the KVM hypervisor.
374 Specifies a list of comma-separated ciphers that SPICE should use
375 for TLS connections. For the format, see man cipher(1).
378 Valid for the KVM hypervisor.
380 Enables or disables passing mouse events via SPICE vdagent.
383 Valid for the Xen HVM and KVM hypervisors.
385 A boolean option that specifies if the hypervisor should enable
386 ACPI support for this instance. By default, ACPI is disabled.
389 Valid for the Xen HVM and KVM hypervisors.
391 A boolean option that specifies if the hypervisor should enabled
392 PAE support for this instance. The default is false, disabling PAE
396 Valid for the Xen HVM and KVM hypervisors.
398 A boolean option that specifies if the instance should be started
399 with its clock set to the localtime of the machine (when true) or
400 to the UTC (When false). The default is false, which is useful for
401 Linux/Unix machines; for Windows OSes, it is recommended to enable
405 Valid for the Xen PVM and KVM hypervisors.
407 This option specifies the path (on the node) to the kernel to boot
408 the instance with. Xen PVM instances always require this, while
409 for KVM if this option is empty, it will cause the machine to load
410 the kernel from its disks.
413 Valid for the Xen PVM and KVM hypervisors.
415 This options specifies extra arguments to the kernel that will be
416 loaded. device. This is always used for Xen PVM, while for KVM it
417 is only used if the ``kernel_path`` option is also specified.
419 The default setting for this value is simply ``"ro"``, which
420 mounts the root disk (initially) in read-only one. For example,
421 setting this to single will cause the instance to start in
425 Valid for the Xen PVM and KVM hypervisors.
427 This option specifies the path (on the node) to the initrd to boot
428 the instance with. Xen PVM instances can use this always, while
429 for KVM if this option is only used if the ``kernel_path`` option
430 is also specified. You can pass here either an absolute filename
431 (the path to the initrd) if you want to use an initrd, or use the
432 format no\_initrd\_path for no initrd.
435 Valid for the Xen PVM and KVM hypervisors.
437 This options specifies the name of the root device. This is always
438 needed for Xen PVM, while for KVM it is only used if the
439 ``kernel_path`` option is also specified.
441 Please note, that if this setting is an empty string and the
442 hypervisor is Xen it will not be written to the Xen configuration
446 Valid for the KVM hypervisor.
448 This boolean option specifies whether to emulate a serial console
452 Valid for the KVM hypervisor.
454 The disk cache mode. It can be either default to not pass any
455 cache option to KVM, or one of the KVM cache modes: none (for
456 direct I/O), writethrough (to use the host cache but report
457 completion to the guest only when the host has committed the
458 changes to disk) or writeback (to use the host cache and report
459 completion as soon as the data is in the host cache). Note that
460 there are special considerations for the cache mode depending on
461 version of KVM used and disk type (always raw file under Ganeti),
462 please refer to the KVM documentation for more details.
465 Valid for the KVM hypervisor.
467 The security model for kvm. Currently one of *none*, *user* or
468 *pool*. Under *none*, the default, nothing is done and instances
469 are run as the Ganeti daemon user (normally root).
471 Under *user* kvm will drop privileges and become the user
472 specified by the security\_domain parameter.
474 Under *pool* a global cluster pool of users will be used, making
475 sure no two instances share the same user on the same node. (this
476 mode is not implemented yet)
479 Valid for the KVM hypervisor.
481 Under security model *user* the username to run the instance
482 under. It must be a valid username existing on the host.
484 Cannot be set under security model *none* or *pool*.
487 Valid for the KVM hypervisor.
489 If *enabled* the -enable-kvm flag is passed to kvm. If *disabled*
490 -disable-kvm is passed. If unset no flag is passed, and the
491 default running mode for your kvm binary will be used.
494 Valid for the KVM hypervisor.
496 This option passes the -mem-path argument to kvm with the path (on
497 the node) to the mount point of the hugetlbfs file system, along
498 with the -mem-prealloc argument too.
501 Valid for the KVM hypervisor.
503 This boolean option determines wether to run the KVM instance in a
506 If it is set to ``true``, an empty directory is created before
507 starting the instance and its path is passed via the -chroot flag
508 to kvm. The directory is removed when the instance is stopped.
510 It is set to ``false`` by default.
513 Valid for the KVM hypervisor.
515 The maximum amount of time (in ms) a KVM instance is allowed to be
516 frozen during a live migration, in order to copy dirty memory
517 pages. Default value is 30ms, but you may need to increase this
518 value for busy instances.
520 This option is only effective with kvm versions >= 87 and qemu-kvm
524 Valid for the LXC hypervisor.
526 The processes belonging to the given instance are only scheduled
527 on the specified CPUs.
529 The parameter format is a comma-separated list of CPU IDs or CPU
530 ID ranges. The ranges are defined by a lower and higher boundary,
531 separated by a dash. The boundaries are inclusive.
534 Valid for the KVM hypervisor.
536 This option specifies the usb mouse type to be used. It can be
537 "mouse" or "tablet". When using VNC it's recommended to set it to
541 Valid for the KVM hypervisor.
543 This option specifies the keyboard mapping to be used. It is only
544 needed when using the VNC console. For example: "fr" or "en-gb".
547 Valid for Xen PVM, Xen HVM and KVM hypervisors.
549 Normally if an instance reboots, the hypervisor will restart it. If
550 this option is set to ``exit``, the hypervisor will treat a reboot
551 as a shutdown instead.
553 It is set to ``reboot`` by default.
556 The ``-O (--os-parameters)`` option allows customisation of the OS
557 parameters. The actual parameter names and values depends on the OS
558 being used, but the syntax is the same key=value. For example, setting
559 a hypothetical ``dhcp`` parameter to yes can be achieved by::
561 gnt-instance add -O dhcp=yes ...
563 The ``-I (--iallocator)`` option specifies the instance allocator
564 plugin to use. If you pass in this option the allocator will select
565 nodes for this instance automatically, so you don't need to pass them
566 with the ``-n`` option. For more information please refer to the
567 instance allocator documentation.
569 The ``-t (--disk-template)`` options specifies the disk layout type
570 for the instance. The available choices are:
573 This creates an instance with no disks. Its useful for testing only
574 (or other special cases).
577 Disk devices will be regular files.
580 Disk devices will be logical volumes.
583 Disk devices will be drbd (version 8.x) on top of lvm volumes.
586 The optional second value of the ``-n (--node)`` is used for the drbd
587 template type and specifies the remote node.
589 If you do not want gnt-instance to wait for the disk mirror to be
590 synced, use the ``--no-wait-for-sync`` option.
592 The ``--file-storage-dir`` specifies the relative path under the
593 cluster-wide file storage directory to store file-based disks. It is
594 useful for having different subdirectories for different
595 instances. The full path of the directory where the disk files are
596 stored will consist of cluster-wide file storage directory + optional
597 subdirectory + instance name. Example:
598 ``@RPL_FILE_STORAGE_DIR@``*/mysubdir/instance1.example.com*. This
599 option is only relevant for instances using the file storage backend.
601 The ``--file-driver`` specifies the driver to use for file-based
602 disks. Note that currently these drivers work with the xen hypervisor
603 only. This option is only relevant for instances using the file
604 storage backend. The available choices are:
607 Kernel loopback driver. This driver uses loopback devices to
608 access the filesystem within the file. However, running I/O
609 intensive applications in your instance using the loop driver
610 might result in slowdowns. Furthermore, if you use the loopback
611 driver consider increasing the maximum amount of loopback devices
612 (on most systems it's 8) using the max\_loop param.
615 The blktap driver (for Xen hypervisors). In order to be able to
616 use the blktap driver you should check if the 'blktapctrl' user
617 space disk agent is running (usually automatically started via
618 xend). This user-level disk I/O interface has the advantage of
619 better performance. Especially if you use a network file system
620 (e.g. NFS) to store your instances this is the recommended choice.
623 The ``--submit`` option is used to send the job to the master daemon
624 but not wait for its completion. The job ID will be shown so that it
625 can be examined via **gnt-job info**.
629 # gnt-instance add -t file --disk 0:size=30g -B maxmem=512 -o debian-etch \
630 -n node1.example.com --file-storage-dir=mysubdir instance1.example.com
631 # gnt-instance add -t plain --disk 0:size=30g -B maxmem=1024,minmem=512 \
632 -o debian-etch -n node1.example.com instance1.example.com
633 # gnt-instance add -t plain --disk 0:size=30g --disk 1:size=100g,vg=san \
634 -B maxmem=512 -o debian-etch -n node1.example.com instance1.example.com
635 # gnt-instance add -t drbd --disk 0:size=30g -B maxmem=512 -o debian-etch \
636 -n node1.example.com:node2.example.com instance2.example.com
642 **batch-create** {instances\_file.json}
644 This command (similar to the Ganeti 1.2 **batcher** tool) submits
645 multiple instance creation jobs based on a definition file. The
646 instance configurations do not encompass all the possible options for
647 the **add** command, but only a subset.
649 The instance file should be a valid-formed JSON file, containing a
650 dictionary with instance name and instance parameters. The accepted
654 The size of the disks of the instance.
657 The disk template to use for the instance, the same as in the
661 A dictionary of backend parameters.
664 A dictionary with a single key (the hypervisor name), and as value
665 the hypervisor options. If not passed, the default hypervisor and
666 hypervisor options will be inherited.
669 Specifications for the one NIC that will be created for the
670 instance. 'bridge' is also accepted as a backwards compatibile
674 List of nics that will be created for the instance. Each entry
675 should be a dict, with mac, ip, mode and link as possible keys.
676 Please don't provide the "mac, ip, mode, link" parent keys if you
677 use this method for specifying nics.
679 primary\_node, secondary\_node
680 The primary and optionally the secondary node to use for the
681 instance (in case an iallocator script is not used).
684 Instead of specifying the nodes, an iallocator script can be used
685 to automatically compute them.
688 whether to start the instance
691 Skip the check for already-in-use instance; see the description in
692 the **add** command for details.
695 Skip the name check for instances; see the description in the
696 **add** command for details.
698 file\_storage\_dir, file\_driver
699 Configuration for the file disk type, see the **add** command for
703 A simple definition for one instance can be (with most of the
704 parameters taken from the cluster defaults)::
710 "disk_size": ["25G"],
716 "disk_size": ["25G"],
717 "iallocator": "dumb",
718 "hypervisor": "xen-hvm",
719 "hvparams": {"acpi": true},
720 "backend": {"maxmem": 512, "minmem": 256}
724 The command will display the job id for each submitted instance, as
727 # gnt-instance batch-create instances.json
734 **remove** [--ignore-failures] [--shutdown-timeout=*N*] [--submit]
735 [--force] {*instance*}
737 Remove an instance. This will remove all data from the instance and
738 there is *no way back*. If you are not sure if you use an instance
739 again, use **shutdown** first and leave it in the shutdown state for a
742 The ``--ignore-failures`` option will cause the removal to proceed
743 even in the presence of errors during the removal of the instance
744 (e.g. during the shutdown or the disk removal). If this option is not
745 given, the command will stop at the first error.
747 The ``--shutdown-timeout`` is used to specify how much time to wait
748 before forcing the shutdown (e.g. ``xm destroy`` in Xen, killing the
749 kvm process for KVM, etc.). By default two minutes are given to each
752 The ``--submit`` option is used to send the job to the master daemon
753 but not wait for its completion. The job ID will be shown so that it
754 can be examined via **gnt-job info**.
756 The ``--force`` option is used to skip the interactive confirmation.
760 # gnt-instance remove instance1.example.com
767 | [--no-headers] [--separator=*SEPARATOR*] [--units=*UNITS*] [-v]
768 | [{-o|--output} *[+]FIELD,...*] [--filter] [instance...]
770 Shows the currently configured instances with memory usage, disk
771 usage, the node they are running on, and their run status.
773 The ``--no-headers`` option will skip the initial header line. The
774 ``--separator`` option takes an argument which denotes what will be
775 used between the output fields. Both these options are to help
778 The units used to display the numeric values in the output varies,
779 depending on the options given. By default, the values will be
780 formatted in the most appropriate unit. If the ``--separator`` option
781 is given, then the values are shown in mebibytes to allow parsing by
782 scripts. In both cases, the ``--units`` option can be used to enforce
785 The ``-v`` option activates verbose mode, which changes the display of
786 special field states (see **ganeti(7)**).
788 The ``-o (--output)`` option takes a comma-separated list of output
789 fields. The available fields and their meaning are:
791 @QUERY_FIELDS_INSTANCE@
793 If the value of the option starts with the character ``+``, the new
794 field(s) will be added to the default list. This allows one to quickly
795 see the default list plus a few other fields, instead of retyping the
796 entire list of fields.
798 There is a subtle grouping about the available output fields: all
799 fields except for ``oper_state``, ``oper_ram``, ``oper_vcpus`` and
800 ``status`` are configuration value and not run-time values. So if you
801 don't select any of the these fields, the query will be satisfied
802 instantly from the cluster configuration, without having to ask the
803 remote nodes for the data. This can be helpful for big clusters when
804 you only want some data and it makes sense to specify a reduced set of
807 If exactly one argument is given and it appears to be a query filter
808 (see **ganeti(7)**), the query result is filtered accordingly. For
809 ambiguous cases (e.g. a single field name as a filter) the ``--filter``
810 (``-F``) option forces the argument to be treated as a filter (e.g.
811 ``gnt-instance list -F admin_state``).
813 The default output field list is: ``name``, ``os``, ``pnode``,
814 ``admin_state``, ``oper_state``, ``oper_ram``.
820 **list-fields** [field...]
822 Lists available fields for instances.
828 **info** [-s \| --static] [--roman] {--all \| *instance*}
830 Show detailed information about the given instance(s). This is
831 different from **list** as it shows detailed data about the instance's
832 disks (especially useful for the drbd disk template).
834 If the option ``-s`` is used, only information available in the
835 configuration file is returned, without querying nodes, making the
838 Use the ``--all`` to get info about all instances, rather than
839 explicitly passing the ones you're interested in.
841 The ``--roman`` option can be used to cause envy among people who like
842 ancient cultures, but are stuck with non-latin-friendly cluster
843 virtualization technologies.
849 | [{-H|--hypervisor-parameters} *HYPERVISOR\_PARAMETERS*]
850 | [{-B|--backend-parameters} *BACKEND\_PARAMETERS*]
851 | [--net add*[:options]* \| --net remove \| --net *N:options*]
852 | [--disk add:size=*SIZE*[,vg=*VG*][,metavg=*VG*] \| --disk remove \|
853 | --disk *N*:mode=*MODE*]
854 | [{-t|--disk-template} plain | {-t|--disk-template} drbd -n *new_secondary*] [--no-wait-for-sync]
855 | [--os-type=*OS* [--force-variant]]
856 | [{-O|--os-parameters} *param*=*value*... ]
857 | [--offline \| --online]
861 Modifies the memory size, number of vcpus, ip address, MAC address
862 and/or nic parameters for an instance. It can also add and remove
863 disks and NICs to/from the instance. Note that you need to give at
864 least one of the arguments, otherwise the command complains.
866 The ``-H (--hypervisor-parameters)``, ``-B (--backend-parameters)``
867 and ``-O (--os-parameters)`` options specifies hypervisor, backend and
868 OS parameter options in the form of name=value[,...]. For details
869 which options can be specified, see the **add** command.
871 The ``-t (--disk-template)`` option will change the disk template of
872 the instance. Currently only conversions between the plain and drbd
873 disk templates are supported, and the instance must be stopped before
874 attempting the conversion. When changing from the plain to the drbd
875 disk template, a new secondary node must be specified via the ``-n``
876 option. The option ``--no-wait-for-sync`` can be used when converting
877 to the ``drbd`` template in order to make the instance available for
878 startup before DRBD has finished resyncing.
880 The ``--disk add:size=``*SIZE* option adds a disk to the instance. The
881 optional ``vg=``*VG* option specifies LVM volume group other than
882 default vg to create the disk on. For DRBD disks, the ``metavg=``*VG*
883 option specifies the volume group for the metadata device. The
884 ``--disk remove`` option will remove the last disk of the
885 instance. The ``--disk`` *N*``:mode=``*MODE* option will change the
886 mode of the Nth disk of the instance between read-only (``ro``) and
889 The ``--net add:``*options* option will add a new NIC to the
890 instance. The available options are the same as in the **add** command
891 (mac, ip, link, mode). The ``--net remove`` will remove the last NIC
892 of the instance, while the ``--net`` *N*:*options* option will change
893 the parameters of the Nth instance NIC.
895 The option ``-o (--os-type)`` will change the OS name for the instance
896 (without reinstallation). In case an OS variant is specified that is
897 not found, then by default the modification is refused, unless
898 ``--force-variant`` is passed. An invalid OS will also be refused,
899 unless the ``--force`` option is given.
901 The ``--online`` and ``--offline`` options are used to transition an
902 instance into and out of the ``offline`` state. An instance can be
903 turned offline only if it was previously down. The ``--online`` option
904 fails if the instance was not in the ``offline`` state, otherwise it
905 changes instance's state to ``down``. These modifications take effect
908 The ``--submit`` option is used to send the job to the master daemon
909 but not wait for its completion. The job ID will be shown so that it
910 can be examined via **gnt-job info**.
912 Most of the changes take effect at the next restart. If the instance is
913 running, there is no effect on the instance.
918 | **reinstall** [{-o|--os-type} *os-type*] [--select-os] [-f *force*]
920 | [--instance \| --node \| --primary \| --secondary \| --all]
921 | [{-O|--os-parameters} *OS\_PARAMETERS*] [--submit] {*instance*...}
923 Reinstalls the operating system on the given instance(s). The
924 instance(s) must be stopped when running this command. If the ``-o
925 (--os-type)`` is specified, the operating system is changed.
927 The ``--select-os`` option switches to an interactive OS reinstall.
928 The user is prompted to select the OS template from the list of
929 available OS templates. OS parameters can be overridden using ``-O
930 (--os-parameters)`` (more documentation for this option under the
933 Since this is a potentially dangerous command, the user will be
934 required to confirm this action, unless the ``-f`` flag is passed.
935 When multiple instances are selected (either by passing multiple
936 arguments or by using the ``--node``, ``--primary``, ``--secondary``
937 or ``--all`` options), the user must pass the ``--force-multiple``
938 options to skip the interactive confirmation.
940 The ``--submit`` option is used to send the job to the master daemon
941 but not wait for its completion. The job ID will be shown so that it
942 can be examined via **gnt-job info**.
947 | **rename** [--no-ip-check] [--no-name-check] [--submit]
948 | {*instance*} {*new\_name*}
950 Renames the given instance. The instance must be stopped when running
951 this command. The requirements for the new name are the same as for
952 adding an instance: the new name must be resolvable and the IP it
953 resolves to must not be reachable (in order to prevent duplicate IPs
954 the next time the instance is started). The IP test can be skipped if
955 the ``--no-ip-check`` option is passed.
957 The ``--no-name-check`` skips the check for the new instance name via
958 the resolver (e.g. in DNS or /etc/hosts, depending on your setup) and
959 that the resolved name matches the provided name. Since the name check
960 is used to compute the IP address, if you pass this option you must also
961 pass the ``--no-ip-check`` option.
963 The ``--submit`` option is used to send the job to the master daemon
964 but not wait for its completion. The job ID will be shown so that it
965 can be examined via **gnt-job info**.
967 Starting/stopping/connecting to console
968 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
974 | [--force] [--ignore-offline]
975 | [--force-multiple] [--no-remember]
976 | [--instance \| --node \| --primary \| --secondary \| --all \|
977 | --tags \| --node-tags \| --pri-node-tags \| --sec-node-tags]
978 | [{-H|--hypervisor-parameters} ``key=value...``]
979 | [{-B|--backend-parameters} ``key=value...``]
980 | [--submit] [--paused]
983 Starts one or more instances, depending on the following options. The
984 four available modes are:
987 will start the instances given as arguments (at least one argument
988 required); this is the default selection
991 will start the instances who have the given node as either primary
995 will start all instances whose primary node is in the list of nodes
996 passed as arguments (at least one node required)
999 will start all instances whose secondary node is in the list of
1000 nodes passed as arguments (at least one node required)
1003 will start all instances in the cluster (no arguments accepted)
1006 will start all instances in the cluster with the tags given as
1010 will start all instances in the cluster on nodes with the tags
1014 will start all instances in the cluster on primary nodes with the
1015 tags given as arguments
1018 will start all instances in the cluster on secondary nodes with the
1019 tags given as arguments
1021 Note that although you can pass more than one selection option, the
1022 last one wins, so in order to guarantee the desired result, don't pass
1023 more than one such option.
1025 Use ``--force`` to start even if secondary disks are failing.
1026 ``--ignore-offline`` can be used to ignore offline primary nodes and
1027 mark the instance as started even if the primary is not available.
1029 The ``--force-multiple`` will skip the interactive confirmation in the
1030 case the more than one instance will be affected.
1032 The ``--no-remember`` option will perform the startup but not change
1033 the state of the instance in the configuration file (if it was stopped
1034 before, Ganeti will still thinks it needs to be stopped). This can be
1035 used for testing, or for a one shot-start where you don't want the
1036 watcher to restart the instance if it crashes.
1038 The ``-H (--hypervisor-parameters)`` and ``-B (--backend-parameters)``
1039 options specify temporary hypervisor and backend parameters that can
1040 be used to start an instance with modified parameters. They can be
1041 useful for quick testing without having to modify an instance back and
1044 # gnt-instance start -H kernel_args="single" instance1
1045 # gnt-instance start -B maxmem=2048 instance2
1048 The first form will start the instance instance1 in single-user mode,
1049 and the instance instance2 with 2GB of RAM (this time only, unless
1050 that is the actual instance memory size already). Note that the values
1051 override the instance parameters (and not extend them): an instance
1052 with "kernel\_args=ro" when started with -H kernel\_args=single will
1053 result in "single", not "ro single". The ``--submit`` option is used
1054 to send the job to the master daemon but not wait for its
1055 completion. The job ID will be shown so that it can be examined via
1058 The ``--paused`` option is only valid for Xen and kvm hypervisors. This
1059 pauses the instance at the start of bootup, awaiting ``gnt-instance
1060 console`` to unpause it, allowing the entire boot process to be
1061 monitored for debugging.
1065 # gnt-instance start instance1.example.com
1066 # gnt-instance start --node node1.example.com node2.example.com
1067 # gnt-instance start --all
1075 | [--force-multiple] [--ignore-offline] [--no-remember]
1076 | [--instance \| --node \| --primary \| --secondary \| --all \|
1077 | --tags \| --node-tags \| --pri-node-tags \| --sec-node-tags]
1081 Stops one or more instances. If the instance cannot be cleanly stopped
1082 during a hardcoded interval (currently 2 minutes), it will forcibly
1083 stop the instance (equivalent to switching off the power on a physical
1086 The ``--timeout`` is used to specify how much time to wait before
1087 forcing the shutdown (e.g. ``xm destroy`` in Xen, killing the kvm
1088 process for KVM, etc.). By default two minutes are given to each
1091 The ``--instance``, ``--node``, ``--primary``, ``--secondary``,
1092 ``--all``, ``--tags``, ``--node-tags``, ``--pri-node-tags`` and
1093 ``--sec-node-tags`` options are similar as for the **startup** command
1094 and they influence the actual instances being shutdown.
1096 The ``--submit`` option is used to send the job to the master daemon
1097 but not wait for its completion. The job ID will be shown so that it
1098 can be examined via **gnt-job info**.
1100 ``--ignore-offline`` can be used to ignore offline primary nodes and
1101 force the instance to be marked as stopped. This option should be used
1102 with care as it can lead to an inconsistent cluster state.
1104 The ``--no-remember`` option will perform the shutdown but not change
1105 the state of the instance in the configuration file (if it was running
1106 before, Ganeti will still thinks it needs to be running). This can be
1107 useful for a cluster-wide shutdown, where some instances are marked as
1108 up and some as down, and you don't want to change the running state:
1109 you just need to disable the watcher, shutdown all instances with
1110 ``--no-remember``, and when the watcher is activated again it will
1111 restore the correct runtime state for all instances.
1115 # gnt-instance shutdown instance1.example.com
1116 # gnt-instance shutdown --all
1123 | [{-t|--type} *REBOOT-TYPE*]
1124 | [--ignore-secondaries]
1125 | [--shutdown-timeout=*N*]
1126 | [--force-multiple]
1127 | [--instance \| --node \| --primary \| --secondary \| --all \|
1128 | --tags \| --node-tags \| --pri-node-tags \| --sec-node-tags]
1132 Reboots one or more instances. The type of reboot depends on the value
1133 of ``-t (--type)``. A soft reboot does a hypervisor reboot, a hard reboot
1134 does a instance stop, recreates the hypervisor config for the instance
1135 and starts the instance. A full reboot does the equivalent of
1136 **gnt-instance shutdown && gnt-instance startup**. The default is
1139 For the hard reboot the option ``--ignore-secondaries`` ignores errors
1140 for the secondary node while re-assembling the instance disks.
1142 The ``--instance``, ``--node``, ``--primary``, ``--secondary``,
1143 ``--all``, ``--tags``, ``--node-tags``, ``--pri-node-tags`` and
1144 ``--sec-node-tags`` options are similar as for the **startup** command
1145 and they influence the actual instances being rebooted.
1147 The ``--shutdown-timeout`` is used to specify how much time to wait
1148 before forcing the shutdown (xm destroy in xen, killing the kvm
1149 process, for kvm). By default two minutes are given to each instance
1152 The ``--force-multiple`` will skip the interactive confirmation in the
1153 case the more than one instance will be affected.
1157 # gnt-instance reboot instance1.example.com
1158 # gnt-instance reboot --type=full instance1.example.com
1164 **console** [--show-cmd] {*instance*}
1166 Connects to the console of the given instance. If the instance is not
1167 up, an error is returned. Use the ``--show-cmd`` option to display the
1168 command instead of executing it.
1170 For HVM instances, this will attempt to connect to the serial console
1171 of the instance. To connect to the virtualized "physical" console of a
1172 HVM instance, use a VNC client with the connection info from the
1175 For Xen/kvm instances, if the instance is paused, this attempts to
1176 unpause the instance after waiting a few seconds for the connection to
1177 the console to be made.
1181 # gnt-instance console instance1.example.com
1190 **replace-disks** [--submit] [--early-release] {-p} [--disks *idx*]
1193 **replace-disks** [--submit] [--early-release] {-s} [--disks *idx*]
1196 **replace-disks** [--submit] [--early-release] {--iallocator *name*
1197 \| --new-secondary *NODE*} {*instance*}
1199 **replace-disks** [--submit] [--early-release] {--auto}
1202 This command is a generalized form for replacing disks. It is
1203 currently only valid for the mirrored (DRBD) disk template.
1205 The first form (when passing the ``-p`` option) will replace the disks
1206 on the primary, while the second form (when passing the ``-s`` option
1207 will replace the disks on the secondary node. For these two cases (as
1208 the node doesn't change), it is possible to only run the replace for a
1209 subset of the disks, using the option ``--disks`` which takes a list
1210 of comma-delimited disk indices (zero-based), e.g. 0,2 to replace only
1211 the first and third disks.
1213 The third form (when passing either the ``--iallocator`` or the
1214 ``--new-secondary`` option) is designed to change secondary node of
1215 the instance. Specifying ``--iallocator`` makes the new secondary be
1216 selected automatically by the specified allocator plugin, otherwise
1217 the new secondary node will be the one chosen manually via the
1218 ``--new-secondary`` option.
1220 The fourth form (when using ``--auto``) will automatically determine
1221 which disks of an instance are faulty and replace them within the same
1222 node. The ``--auto`` option works only when an instance has only
1223 faulty disks on either the primary or secondary node; it doesn't work
1224 when both sides have faulty disks.
1226 The ``--submit`` option is used to send the job to the master daemon
1227 but not wait for its completion. The job ID will be shown so that it
1228 can be examined via **gnt-job info**.
1230 The ``--early-release`` changes the code so that the old storage on
1231 secondary node(s) is removed early (before the resync is completed)
1232 and the internal Ganeti locks for the current (and new, if any)
1233 secondary node are also released, thus allowing more parallelism in
1234 the cluster operation. This should be used only when recovering from a
1235 disk failure on the current secondary (thus the old storage is already
1236 broken) or when the storage on the primary node is known to be fine
1237 (thus we won't need the old storage for potential recovery).
1239 Note that it is not possible to select an offline or drained node as a
1245 **activate-disks** [--submit] [--ignore-size] {*instance*}
1247 Activates the block devices of the given instance. If successful, the
1248 command will show the location and name of the block devices::
1250 node1.example.com:disk/0:/dev/drbd0
1251 node1.example.com:disk/1:/dev/drbd1
1254 In this example, *node1.example.com* is the name of the node on which
1255 the devices have been activated. The *disk/0* and *disk/1* are the
1256 Ganeti-names of the instance disks; how they are visible inside the
1257 instance is hypervisor-specific. */dev/drbd0* and */dev/drbd1* are the
1258 actual block devices as visible on the node. The ``--submit`` option
1259 is used to send the job to the master daemon but not wait for its
1260 completion. The job ID will be shown so that it can be examined via
1263 The ``--ignore-size`` option can be used to activate disks ignoring
1264 the currently configured size in Ganeti. This can be used in cases
1265 where the configuration has gotten out of sync with the real-world
1266 (e.g. after a partially-failed grow-disk operation or due to rounding
1267 in LVM devices). This should not be used in normal cases, but only
1268 when activate-disks fails without it.
1270 Note that it is safe to run this command while the instance is already
1276 **deactivate-disks** [-f] [--submit] {*instance*}
1278 De-activates the block devices of the given instance. Note that if you
1279 run this command for an instance with a drbd disk template, while it
1280 is running, it will not be able to shutdown the block devices on the
1281 primary node, but it will shutdown the block devices on the secondary
1282 nodes, thus breaking the replication.
1284 The ``-f``/``--force`` option will skip checks that the instance is
1285 down; in case the hypervisor is confused and we can't talk to it,
1286 normally Ganeti will refuse to deactivate the disks, but with this
1287 option passed it will skip this check and directly try to deactivate
1288 the disks. This can still fail due to the instance actually running or
1291 The ``--submit`` option is used to send the job to the master daemon
1292 but not wait for its completion. The job ID will be shown so that it
1293 can be examined via **gnt-job info**.
1298 **grow-disk** [--no-wait-for-sync] [--submit] {*instance*} {*disk*}
1301 Grows an instance's disk. This is only possible for instances having a
1302 plain or drbd disk template.
1304 Note that this command only change the block device size; it will not
1305 grow the actual filesystems, partitions, etc. that live on that
1306 disk. Usually, you will need to:
1308 #. use **gnt-instance grow-disk**
1310 #. reboot the instance (later, at a convenient time)
1312 #. use a filesystem resizer, such as ext2online(8) or
1313 xfs\_growfs(8) to resize the filesystem, or use fdisk(8) to change
1314 the partition table on the disk
1316 The *disk* argument is the index of the instance disk to grow. The
1317 *amount* argument is given either as a number (and it represents the
1318 amount to increase the disk with in mebibytes) or can be given similar
1319 to the arguments in the create instance operation, with a suffix
1322 Note that the disk grow operation might complete on one node but fail
1323 on the other; this will leave the instance with different-sized LVs on
1324 the two nodes, but this will not create problems (except for unused
1327 If you do not want gnt-instance to wait for the new disk region to be
1328 synced, use the ``--no-wait-for-sync`` option.
1330 The ``--submit`` option is used to send the job to the master daemon
1331 but not wait for its completion. The job ID will be shown so that it
1332 can be examined via **gnt-job info**.
1334 Example (increase the first disk for instance1 by 16GiB)::
1336 # gnt-instance grow-disk instance1.example.com 0 16g
1339 Also note that disk shrinking is not supported; use **gnt-backup
1340 export** and then **gnt-backup import** to reduce the disk size of an
1346 **recreate-disks** [--submit] [--disks=``indices``] [-n node1:[node2]]
1349 Recreates the disks of the given instance, or only a subset of the
1350 disks (if the option ``disks`` is passed, which must be a
1351 comma-separated list of disk indices, starting from zero).
1353 Note that this functionality should only be used for missing disks; if
1354 any of the given disks already exists, the operation will fail. While
1355 this is suboptimal, recreate-disks should hopefully not be needed in
1356 normal operation and as such the impact of this is low.
1358 Optionally the instance's disks can be recreated on different
1359 nodes. This can be useful if, for example, the original nodes of the
1360 instance have gone down (and are marked offline), so we can't recreate
1361 on the same nodes. To do this, pass the new node(s) via ``-n`` option,
1362 with a syntax similar to the **add** command. The number of nodes
1363 passed must equal the number of nodes that the instance currently
1364 has. Note that changing nodes is only allowed for 'all disk'
1365 replacement (when ``--disks`` is not passed).
1367 The ``--submit`` option is used to send the job to the master daemon
1368 but not wait for its completion. The job ID will be shown so that it
1369 can be examined via **gnt-job info**.
1377 **failover** [-f] [--ignore-consistency] [--shutdown-timeout=*N*]
1378 [--submit] {*instance*}
1380 Failover will stop the instance (if running), change its primary node,
1381 and if it was originally running it will start it again (on the new
1382 primary). This only works for instances with drbd template (in which
1383 case you can only fail to the secondary node) and for externally
1384 mirrored templates (shared storage) (which can change to any other
1387 Normally the failover will check the consistency of the disks before
1388 failing over the instance. If you are trying to migrate instances off
1389 a dead node, this will fail. Use the ``--ignore-consistency`` option
1390 for this purpose. Note that this option can be dangerous as errors in
1391 shutting down the instance will be ignored, resulting in possibly
1392 having the instance running on two machines in parallel (on
1393 disconnected DRBD drives).
1395 The ``--shutdown-timeout`` is used to specify how much time to wait
1396 before forcing the shutdown (xm destroy in xen, killing the kvm
1397 process, for kvm). By default two minutes are given to each instance
1400 The ``--submit`` option is used to send the job to the master daemon
1401 but not wait for its completion. The job ID will be shown so that it
1402 can be examined via **gnt-job info**.
1406 # gnt-instance failover instance1.example.com
1412 **migrate** [-f] {--cleanup} {*instance*}
1414 **migrate** [-f] [--allow-failover] [--non-live]
1415 [--migration-mode=live\|non-live] {*instance*}
1417 Migrate will move the instance to its secondary node without
1418 shutdown. It only works for instances having the drbd8 disk template
1421 The migration command needs a perfectly healthy instance, as we rely
1422 on the dual-master capability of drbd8 and the disks of the instance
1423 are not allowed to be degraded.
1425 The ``--non-live`` and ``--migration-mode=non-live`` options will
1426 switch (for the hypervisors that support it) between a "fully live"
1427 (i.e. the interruption is as minimal as possible) migration and one in
1428 which the instance is frozen, its state saved and transported to the
1429 remote node, and then resumed there. This all depends on the
1430 hypervisor support for two different methods. In any case, it is not
1431 an error to pass this parameter (it will just be ignored if the
1432 hypervisor doesn't support it). The option ``--migration-mode=live``
1433 option will request a fully-live migration. The default, when neither
1434 option is passed, depends on the hypervisor parameters (and can be
1435 viewed with the **gnt-cluster info** command).
1437 If the ``--cleanup`` option is passed, the operation changes from
1438 migration to attempting recovery from a failed previous migration. In
1439 this mode, Ganeti checks if the instance runs on the correct node (and
1440 updates its configuration if not) and ensures the instances's disks
1441 are configured correctly. In this mode, the ``--non-live`` option is
1444 The option ``-f`` will skip the prompting for confirmation.
1446 If ``--allow-failover`` is specified it tries to fallback to failover if
1447 it already can determine that a migration won't work (i.e. if the
1448 instance is shut down). Please note that the fallback will not happen
1449 during execution. If a migration fails during execution it still fails.
1451 Example (and expected output)::
1453 # gnt-instance migrate instance1
1454 Instance instance1 will be migrated. Note that migration
1455 might impact the instance if anything goes wrong (e.g. due to bugs in
1456 the hypervisor). Continue?
1458 Migrating instance instance1.example.com
1459 * checking disk consistency between source and target
1460 * switching node node2.example.com to secondary mode
1461 * changing into standalone mode
1462 * changing disks into dual-master mode
1463 * wait until resync is done
1464 * preparing node2.example.com to accept the instance
1465 * migrating instance to node2.example.com
1466 * switching node node1.example.com to secondary mode
1467 * wait until resync is done
1468 * changing into standalone mode
1469 * changing disks into single-master mode
1470 * wait until resync is done
1478 **move** [-f] [--ignore-consistency]
1479 [-n *node*] [--shutdown-timeout=*N*] [--submit]
1482 Move will move the instance to an arbitrary node in the cluster. This
1483 works only for instances having a plain or file disk template.
1485 Note that since this operation is done via data copy, it will take a
1486 long time for big disks (similar to replace-disks for a drbd
1489 The ``--shutdown-timeout`` is used to specify how much time to wait
1490 before forcing the shutdown (e.g. ``xm destroy`` in XEN, killing the
1491 kvm process for KVM, etc.). By default two minutes are given to each
1494 The ``--ignore-consistency`` option will make Ganeti ignore any errors
1495 in trying to shutdown the instance on its node; useful if the
1496 hypervisor is broken and you want to recuperate the data.
1498 The ``--submit`` option is used to send the job to the master daemon
1499 but not wait for its completion. The job ID will be shown so that it
1500 can be examined via **gnt-job info**.
1504 # gnt-instance move -n node3.example.com instance1.example.com
1510 **change-group** [--iallocator *NAME*] [--to *GROUP*...] {*instance*}
1512 This command moves an instance to another node group. The move is
1513 calculated by an iallocator, either given on the command line or as a
1516 If no specific destination groups are specified using ``--to``, all
1517 groups except the one containing the instance are considered.
1521 # gnt-instance change-group -I hail --to rack2 inst1.example.com
1530 **add-tags** [--from *file*] {*instancename*} {*tag*...}
1532 Add tags to the given instance. If any of the tags contains invalid
1533 characters, the entire operation will abort.
1535 If the ``--from`` option is given, the list of tags will be extended
1536 with the contents of that file (each line becomes a tag). In this
1537 case, there is not need to pass tags on the command line (if you do,
1538 both sources will be used). A file name of ``-`` will be interpreted
1544 **list-tags** {*instancename*}
1546 List the tags of the given instance.
1551 **remove-tags** [--from *file*] {*instancename*} {*tag*...}
1553 Remove tags from the given instance. If any of the tags are not
1554 existing on the node, the entire operation will abort.
1556 If the ``--from`` option is given, the list of tags to be removed will
1557 be extended with the contents of that file (each line becomes a tag).
1558 In this case, there is not need to pass tags on the command line (if
1559 you do, tags from both sources will be removed). A file name of ``-``
1560 will be interpreted as stdin.
1562 .. vim: set textwidth=72 :