1 <!doctype refentry PUBLIC "-//OASIS//DTD DocBook V4.1//EN" [
3 <!-- Fill in your name for FIRSTNAME and SURNAME. -->
4 <!-- Please adjust the date whenever revising the manpage. -->
5 <!ENTITY dhdate "<date>June 08, 2010</date>">
6 <!-- SECTION should be 1-8, maybe w/ subsection other parameters are
7 allowed: see man(7), man(1). -->
8 <!ENTITY dhsection "<manvolnum>8</manvolnum>">
9 <!ENTITY dhucpackage "<refentrytitle>gnt-instance</refentrytitle>">
10 <!ENTITY dhpackage "gnt-instance">
12 <!ENTITY debian "<productname>Debian</productname>">
13 <!ENTITY gnu "<acronym>GNU</acronym>">
14 <!ENTITY gpl "&gnu; <acronym>GPL</acronym>">
15 <!ENTITY footer SYSTEM "footer.sgml">
26 <holder>Google Inc.</holder>
34 <refmiscinfo>Ganeti 2.2</refmiscinfo>
37 <refname>&dhpackage;</refname>
39 <refpurpose>Ganeti instance administration</refpurpose>
43 <command>&dhpackage; </command>
45 <arg choice="req">command</arg>
46 <arg>arguments...</arg>
50 <title>DESCRIPTION</title>
53 The <command>&dhpackage;</command> is used for instance
54 administration in the Ganeti system.
59 <title>COMMANDS</title>
62 <title>Creation/removal/querying</title>
67 <command>add</command>
69 <arg choice="req">-t<group choice="req">
78 <arg rep="repeat">--disk=<replaceable>N</replaceable>:<group choice="req">
79 <arg>size=<replaceable>VAL</replaceable></arg>
80 <arg>adopt=<replaceable>LV</replaceable></arg>
81 </group>,mode=<replaceable>ro|rw</replaceable></arg>
82 <arg>-s <replaceable>SIZE</replaceable></arg>
85 <arg>--no-ip-check</arg>
86 <arg>--no-name-check</arg>
88 <arg>--no-install</arg>
91 <arg rep="repeat">--net=<replaceable>N</replaceable><arg rep="repeat">:options</arg></arg>
95 <arg>-B <replaceable>BEPARAMS</replaceable></arg>
98 <arg>-H <replaceable>HYPERVISOR</replaceable><arg>:<arg choice="plain" rep="repeat">option=<replaceable>value</replaceable></arg></arg></arg>
101 <arg>--file-storage-dir <replaceable>dir_path</replaceable></arg>
102 <arg>--file-driver<group choice="req">
109 <arg>-n <replaceable>node<optional>:secondary-node</optional></replaceable></arg>
110 <arg>--iallocator <replaceable>name</replaceable></arg>
114 <arg choice="req">-o <replaceable>os-type</replaceable></arg>
119 <arg choice="req"><replaceable>instance</replaceable></arg>
123 Creates a new instance on the specified host. The
124 <replaceable>instance</replaceable> argument must be in DNS,
125 but depending on the bridge/routing setup, need not be in
126 the same network as the nodes in the cluster.
130 The <option>disk</option> option specifies the parameters
131 for the disks of the instance. The numbering of disks starts
132 at zero, and at least one disk needs to be passed. For each
133 disk, either the size or the adoption source needs to be
134 given, and optionally the access mode (read-only or the
135 default of read-write) can also be specified. The size is
136 interpreted (when no unit is given) in mebibytes. You can
137 also use one of the suffixes
138 <literal>m</literal>, <literal>g</literal> or
139 <literal>t</literal> to specify the exact the units used;
140 these suffixes map to mebibytes, gibibytes and tebibytes.
144 When using the <option>adopt</option> key in the disk
145 definition, Ganeti will reuse those volumes (instead of
146 creating new ones) as the instance's disks. Ganeti will
147 rename these volumes to the standard format, and (without
148 installing the OS) will use them as-is for the
149 instance. This allows migrating instances from non-managed
150 mode (e.q. plain KVM with LVM) to being managed via
151 Ganeti. Note that this works only for the `plain' disk
152 template (see below for template details).
156 Alternatively, a single-disk instance can be created via the
157 <option>-s</option> option which takes a single argument,
158 the size of the disk. This is similar to the Ganeti 1.2
159 version (but will only create one disk).
163 The minimum disk specification is therefore
164 <userinput>--disk 0:size=20G</userinput> (or <userinput>-s
165 20G</userinput> when using the <option>-s</option> option),
166 and a three-disk instance can be specified as
167 <userinput>--disk 0:size=20G --disk 1:size=4G --disk
168 2:size=100G</userinput>.
172 The <option>--no-ip-check</option> skips the checks that are
173 done to see if the instance's IP is not already alive
174 (i.e. reachable from the master node).
178 The <option>--no-name-check</option> skips the check for the
179 instance name via the resolver (e.g. in DNS or /etc/hosts,
180 depending on your setup). Since the name check is used to
181 compute the IP address, if you pass this option you must
182 also pass the <option>--no-ip-check</option> option.
186 If you don't wat the instance to automatically start after
187 creation, this is possible via the
188 <option>--no-start</option> option. This will leave the
189 instance down until a subsequent <command>gnt-instance
190 start</command> command.
194 The NICs of the instances can be specified via the
195 <option>--net</option> option. By default, one NIC is
196 created for the instance, with a random MAC, and set
197 up according the the cluster level nic parameters.
198 Each NIC can take these parameters (all optional):
203 <simpara>either a value or 'generate' to generate a
204 new unique MAC</simpara>
210 <simpara>specifies the IP address assigned to the
211 instance from the Ganeti side (this is not necessarily
212 what the instance will use, but what the node expects
213 the instance to use)</simpara>
219 <simpara>specifies the connection mode for this nic:
220 routed or bridged.</simpara>
226 <simpara>in bridged mode specifies the bridge to attach
227 this NIC to, in routed mode it's intended to
228 differentiate between different routing tables/instance
229 groups (but the meaning is dependent on the network
230 script, see gnt-cluster(8) for more details)</simpara>
234 Of these "mode" and "link" are nic parameters, and inherit their
235 default at cluster level.
239 Alternatively, if no network is desired for the instance, you
240 can prevent the default of one NIC with the
241 <option>--no-nics</option> option.
245 The <option>-o</option> options specifies the operating
246 system to be installed. The available operating systems can
247 be listed with <command>gnt-os
248 list</command>. Passing <option>--no-install</option> will
249 however skip the OS installation, allowing a manual import
250 if so desired. Note that the no-installation mode will
251 automatically disable the start-up of the instance (without
252 an OS, it most likely won't be able to start-up
257 The <option>-B</option> option specifies the backend
258 parameters for the instance. If no such parameters are
259 specified, the values are inherited from the cluster. Possible
265 <simpara>the memory size of the instance; as usual,
266 suffixes can be used to denote the unit, otherwise the
267 value is taken in mebibites</simpara>
273 <simpara>the number of VCPUs to assign to the instance
274 (if this value makes sense for the hypervisor)</simpara>
278 <term>auto_balance</term>
280 <simpara>whether the instance is considered in the N+1
281 cluster checks (enough redundancy in the cluster to
282 survive a node failure)</simpara>
289 The <option>-H</option> option specified the hypervisor to
290 use for the instance (must be one of the enabled hypervisors
291 on the cluster) and optionally custom parameters for this
292 instance. If not other options are used (i.e. the invocation
293 is just <userinput>-H
294 <replaceable>NAME</replaceable></userinput>) the instance
295 will inherit the cluster options. The defaults below show
296 the cluster defaults at cluster creation time.
300 The possible hypervisor options are as follows:
303 <term>boot_order</term>
305 <simpara>Valid for the Xen HVM and KVM
306 hypervisors.</simpara>
308 <simpara>A string value denoting the boot order. This
309 has different meaning for the Xen HVM hypervisor and
310 for the KVM one.</simpara>
313 For Xen HVM, The boot order is a string of letters
314 listing the boot devices, with valid device letters
352 The default is not to set an HVM boot order which is
357 For KVM the boot order is either
358 <quote>cdrom</quote>, <quote>disk</quote> or
359 <quote>network</quote>. Please note that older
360 versions of KVM couldn't netboot from virtio
361 interfaces. This has been fixed in more recent
362 versions and is confirmed to work at least with
369 <term>cdrom_image_path</term>
371 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
373 <simpara>The path to a CDROM image to attach to the
379 <term>nic_type</term>
381 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
384 This parameter determines the way the network cards
385 are presented to the instance. The possible options are:
387 <member>rtl8139 (default for Xen HVM) (HVM & KVM)</member>
388 <member>ne2k_isa (HVM & KVM)</member>
389 <member>ne2k_pci (HVM & KVM)</member>
390 <member>i82551 (KVM)</member>
391 <member>i82557b (KVM)</member>
392 <member>i82559er (KVM)</member>
393 <member>pcnet (KVM)</member>
394 <member>e1000 (KVM)</member>
395 <member>paravirtual (default for KVM) (HVM & KVM)</member>
401 <term>disk_type</term>
403 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
406 This parameter determines the way the disks are
407 presented to the instance. The possible options are:
409 <member>ioemu (default for HVM & KVM) (HVM & KVM)</member>
410 <member>ide (HVM & KVM)</member>
411 <member>scsi (KVM)</member>
412 <member>sd (KVM)</member>
413 <member>mtd (KVM)</member>
414 <member>pflash (KVM)</member>
420 <term>vnc_bind_address</term>
422 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
424 <para>Specifies the address that the VNC listener for
425 this instance should bind to. Valid values are IPv4
426 addresses. Use the address 0.0.0.0 to bind to all
427 available interfaces (this is the default) or specify
428 the address of one of the interfaces on the node to
429 restrict listening to that interface.</para>
436 <simpara>Valid for the KVM hypervisor.</simpara>
438 <simpara>A boolean option that controls whether the
439 VNC connection is secured with TLS.</simpara>
444 <term>vnc_x509_path</term>
446 <simpara>Valid for the KVM hypervisor.</simpara>
448 <para>If <option>vnc_tls</option> is enabled, this
449 options specifies the path to the x509 certificate to
455 <term>vnc_x509_verify</term>
457 <simpara>Valid for the KVM hypervisor.</simpara>
464 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
467 A boolean option that specifies if the hypervisor
468 should enable ACPI support for this instance. By
469 default, ACPI is disabled.
477 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
480 A boolean option that specifies if the hypervisor
481 should enabled PAE support for this instance. The
482 default is false, disabling PAE support.
488 <term>use_localtime</term>
490 <simpara>Valid for the Xen HVM and KVM hypervisors.</simpara>
493 A boolean option that specifies if the instance
494 should be started with its clock set to the
495 localtime of the machine (when true) or to the UTC
496 (When false). The default is false, which is useful
497 for Linux/Unix machines; for Windows OSes, it is
498 recommended to enable this parameter.
504 <term>kernel_path</term>
506 <simpara>Valid for the Xen PVM and KVM hypervisors.</simpara>
509 This option specifies the path (on the node) to the
510 kernel to boot the instance with. Xen PVM instances
511 always require this, while for KVM if this option is
512 empty, it will cause the machine to load the kernel
519 <term>kernel_args</term>
521 <simpara>Valid for the Xen PVM and KVM hypervisors.</simpara>
524 This options specifies extra arguments to the kernel
525 that will be loaded. device. This is always used
526 for Xen PVM, while for KVM it is only used if the
527 <option>kernel_path</option> option is also
532 The default setting for this value is simply
533 <constant>"ro"</constant>, which mounts the root
534 disk (initially) in read-only one. For example,
535 setting this to <userinput>single</userinput> will
536 cause the instance to start in single-user mode.
542 <term>initrd_path</term>
544 <simpara>Valid for the Xen PVM and KVM hypervisors.</simpara>
547 This option specifies the path (on the node) to the
548 initrd to boot the instance with. Xen PVM instances
549 can use this always, while for KVM if this option is
550 only used if the <option>kernel_path</option> option
551 is also specified. You can pass here either an
552 absolute filename (the path to the initrd) if you
553 want to use an initrd, or use the format
554 <userinput>no_initrd_path</userinput> for no initrd.
560 <term>root_path</term>
562 <simpara>Valid for the Xen PVM and KVM hypervisors.</simpara>
565 This options specifies the name of the root
566 device. This is always needed for Xen PVM, while for
567 KVM it is only used if the
568 <option>kernel_path</option> option is also
575 <term>serial_console</term>
577 <simpara>Valid for the KVM hypervisor.</simpara>
579 <simpara>This boolean option specifies whether to
580 emulate a serial console for the instance.</simpara>
585 <term>disk_cache</term>
587 <simpara>Valid for the KVM hypervisor.</simpara>
589 <simpara>The disk cache mode. It can be either
590 <userinput>default</userinput> to not pass any cache
591 option to KVM, or one of the KVM cache modes: none
592 (for direct I/O), writethrough (to use the host cache
593 but report completion to the guest only when the host
594 has committed the changes to disk) or writeback (to
595 use the host cache and report completion as soon as
596 the data is in the host cache). Note that there are
597 special considerations for the cache mode depending on
598 version of KVM used and disk type (always raw file
599 under Ganeti), please refer to the KVM documentation
606 <term>security_model</term>
608 <simpara>Valid for the KVM hypervisor.</simpara>
610 <simpara>The security model for kvm. Currently one of
611 <quote>none</quote>, <quote>user</quote> or
612 <quote>pool</quote>. Under <quote>none</quote>, the
613 default, nothing is done and instances are run as
614 the Ganeti daemon user (normally root).
617 <simpara>Under <quote>user</quote> kvm will drop
618 privileges and become the user specified by the
619 security_domain parameter.
622 <simpara>Under <quote>pool</quote> a global cluster
623 pool of users will be used, making sure no two
624 instances share the same user on the same node.
625 (this mode is not implemented yet)
632 <term>security_domain</term>
634 <simpara>Valid for the KVM hypervisor.</simpara>
636 <simpara>Under security model <quote>user</quote> the
637 username to run the instance under. It must be a valid
638 username existing on the host.
640 <simpara>Cannot be set under security model <quote>none</quote>
641 or <quote>pool</quote>.
648 <term>kvm_flag</term>
650 <simpara>Valid for the KVM hypervisor.</simpara>
652 <simpara>If <quote>enabled</quote> the -enable-kvm flag is
653 passed to kvm. If <quote>disabled</quote> -disable-kvm is
654 passed. If unset no flag is passed, and the default running
655 mode for your kvm binary will be used.
662 <term>mem_path</term>
664 <simpara>Valid for the KVM hypervisor.</simpara>
666 <simpara>This option passes the -mem-path argument to kvm with
667 the path (on the node) to the mount point of the hugetlbfs
668 file system, along with the -mem-prealloc argument too.
675 <term>use_chroot</term>
677 <simpara>Valid for the KVM hypervisor.</simpara>
679 <simpara>This boolean option determines wether to run the KVM
680 instance in a chroot directory.
682 <para>If it is set to <quote>true</quote>, an empty directory
683 is created before starting the instance and its path is passed via
684 the -chroot flag to kvm.
685 The directory is removed when the instance is stopped.
688 <simpara>It is set to <quote>false</quote> by default.</simpara>
693 <term>migration_downtime</term>
695 <simpara>Valid for the KVM hypervisor.</simpara>
697 <simpara>The maximum amount of time (in ms) a KVM instance is
698 allowed to be frozen during a live migration, in order to copy
699 dirty memory pages. Default value is 30ms, but you may need to
700 increase this value for busy instances.
703 <simpara>This option is only effective with kvm versions >= 87
704 and qemu-kvm versions >= 0.11.0.
710 <term>use_chroot</term>
712 <simpara>Valid for the KVM hypervisor.</simpara>
714 <simpara>This boolean option determines wether to run the KVM
715 instance in a chroot directory.
718 <para>If it is set to <quote>true</quote>, an empty
719 directory is created before starting the instance and
720 its path is passed via the <option>-chroot</option>
721 flag to kvm. The directory is removed when the
725 <simpara>It is set to <quote>false</quote> by
732 <term>cpu_mask</term>
734 <simpara>Valid for the LXC hypervisor.</simpara>
736 <simpara>The processes belonging to the given instance are
737 only scheduled on the specified CPUs.
741 The parameter format is a comma-separated list of CPU IDs or
742 CPU ID ranges. The ranges are defined by a lower and higher
743 boundary, separated by a dash. The boundaries are inclusive.
750 <term>usb_mouse</term>
752 <simpara>Valid for the KVM hypervisor.</simpara>
754 <simpara>This option specifies the usb mouse type to be used.
755 It can be <quote>mouse</quote> or <quote>tablet</quote>. When
756 using VNC it's recommended to set it to <quote>tablet</quote>.
766 The <option>--iallocator</option> option specifies the instance
767 allocator plugin to use. If you pass in this option the allocator
768 will select nodes for this instance automatically, so you don't need
769 to pass them with the <option>-n</option> option. For more
770 information please refer to the instance allocator documentation.
774 The <option>-t</option> options specifies the disk layout type for
775 the instance. The available choices are:
778 <term>diskless</term>
781 This creates an instance with no disks. Its useful for
782 testing only (or other special cases).
789 <para>Disk devices will be regular files.</para>
795 <para>Disk devices will be logical volumes.</para>
802 Disk devices will be drbd (version 8.x) on top of
811 The optional second value of the <option>--node</option> is used for
812 the drbd template type and specifies the remote node.
816 If you do not want gnt-instance to wait for the disk mirror
817 to be synced, use the <option>--no-wait-for-sync</option>
822 The <option>--file-storage-dir</option> specifies the relative path
823 under the cluster-wide file storage directory to store file-based
824 disks. It is useful for having different subdirectories for
825 different instances. The full path of the directory where the disk
826 files are stored will consist of cluster-wide file storage directory
827 + optional subdirectory + instance name. Example:
828 /srv/ganeti/file-storage/mysubdir/instance1.example.com. This option
829 is only relevant for instances using the file storage backend.
833 The <option>--file-driver</option> specifies the driver to use for
834 file-based disks. Note that currently these drivers work with the
835 xen hypervisor only. This option is only relevant for instances using
836 the file storage backend. The available choices are:
842 Kernel loopback driver. This driver uses loopback
843 devices to access the filesystem within the
844 file. However, running I/O intensive applications in
845 your instance using the loop driver might result in
846 slowdowns. Furthermore, if you use the loopback
847 driver consider increasing the maximum amount of
848 loopback devices (on most systems it's 8) using the
856 <para>The blktap driver (for Xen hypervisors). In
857 order to be able to use the blktap driver you should
858 check if the 'blktapctrl' user space disk agent is
859 running (usually automatically started via xend). This
860 user-level disk I/O interface has the advantage of
861 better performance. Especially if you use a network
862 file system (e.g. NFS) to store your instances this is
863 the recommended choice.
871 The <option>--submit</option> option is used to send the job to
872 the master daemon but not wait for its completion. The job
873 ID will be shown so that it can be examined via
874 <command>gnt-job info</command>.
880 # gnt-instance add -t file --disk 0:size=30g -B memory=512 -o debian-etch \
881 -n node1.example.com --file-storage-dir=mysubdir instance1.example.com
882 # gnt-instance add -t plain --disk 0:size=30g -B memory=512 -o debian-etch \
883 -n node1.example.com instance1.example.com
884 # gnt-instance add -t drbd --disk 0:size=30g -B memory=512 -o debian-etch \
885 -n node1.example.com:node2.example.com instance2.example.com
891 <title>BATCH-CREATE</title>
893 <command>batch-create</command>
894 <arg choice="req">instances_file.json</arg>
898 This command (similar to the Ganeti 1.2
899 <command>batcher</command> tool) submits multiple instance
900 creation jobs based on a definition file. The instance
901 configurations do not encompass all the possible options for
902 the <command>add</command> command, but only a subset.
906 The instance file should be a valid-formed JSON file,
907 containing a dictionary with instance name and instance
908 parameters. The accepted parameters are:
912 <term>disk_size</term>
914 <simpara>The size of the disks of the instance.</simpara>
918 <term>disk_template</term>
920 <simpara>The disk template to use for the instance,
921 the same as in the <command>add</command>
928 <simpara>A dictionary of backend parameters.</simpara>
932 <term>hypervisor</term>
934 <simpara>A dictionary with a single key (the
935 hypervisor name), and as value the hypervisor
936 options. If not passed, the default hypervisor and
937 hypervisor options will be inherited.</simpara>
941 <term>mac, ip, mode, link</term>
943 <simpara>Specifications for the one NIC that will be
944 created for the instance. 'bridge' is also accepted
945 as a backwards compatibile key.</simpara>
951 <simpara>List of nics that will be created for the
952 instance. Each entry should be a dict, with mac, ip, mode
953 and link as possible keys. Please don't provide the "mac,
954 ip, mode, link" parent keys if you use this method for
955 specifying nics.</simpara>
959 <term>primary_node, secondary_node</term>
961 <simpara>The primary and optionally the secondary node
962 to use for the instance (in case an iallocator script
963 is not used).</simpara>
967 <term>iallocator</term>
969 <simpara>Instead of specifying the nodes, an
970 iallocator script can be used to automatically compute
977 <simpara>whether to start the instance</simpara>
981 <term>ip_check</term>
983 <simpara>Skip the check for already-in-use instance;
984 see the description in the <command>add</command>
985 command for details.</simpara>
989 <term>name_check</term>
991 <simpara>Skip the name check for instances;
992 see the description in the <command>add</command>
993 command for details.</simpara>
997 <term>file_storage_dir, file_driver</term>
999 <simpara>Configuration for the <literal>file</literal>
1000 disk type, see the <command>add</command> command for
1008 A simple definition for one instance can be (with most of
1009 the parameters taken from the cluster defaults):
1014 "os": "debootstrap",
1015 "disk_size": ["25G"],
1016 "iallocator": "dumb"
1020 "os": "debootstrap",
1021 "disk_size": ["25G"],
1022 "iallocator": "dumb",
1023 "hypervisor": "xen-hvm",
1024 "hvparams": {"acpi": true},
1025 "backend": {"memory": 512}
1032 The command will display the job id for each submitted
1033 instance, as follows:
1035 # gnt-instance batch-create instances.json
1044 <title>REMOVE</title>
1047 <command>remove</command>
1048 <arg>--ignore-failures</arg>
1049 <arg>--shutdown-timeout=<replaceable>N</replaceable></arg>
1051 <arg choice="req"><replaceable>instance</replaceable></arg>
1055 Remove an instance. This will remove all data from the
1056 instance and there is <emphasis>no way back</emphasis>. If
1057 you are not sure if you use an instance again, use
1058 <command>shutdown</command> first and leave it in the
1059 shutdown state for a while.
1064 The <option>--ignore-failures</option> option will cause the
1065 removal to proceed even in the presence of errors during the
1066 removal of the instance (e.g. during the shutdown or the
1067 disk removal). If this option is not given, the command will
1068 stop at the first error.
1072 The <option>--shutdown-timeout</option> is used to specify how
1073 much time to wait before forcing the shutdown (xm destroy in xen,
1074 killing the kvm process, for kvm). By default two minutes are
1075 given to each instance to stop.
1079 The <option>--submit</option> option is used to send the job to
1080 the master daemon but not wait for its completion. The job
1081 ID will be shown so that it can be examined via
1082 <command>gnt-job info</command>.
1088 # gnt-instance remove instance1.example.com
1097 <command>list</command>
1098 <arg>--no-headers</arg>
1099 <arg>--separator=<replaceable>SEPARATOR</replaceable></arg>
1101 <arg>--units=<replaceable>UNITS</replaceable></arg>
1102 <arg>-o <replaceable>[+]FIELD,...</replaceable></arg>
1104 <arg rep="repeat">instance</arg>
1108 Shows the currently configured instances with memory usage,
1109 disk usage, the node they are running on, and their run
1114 The <option>--no-headers</option> option will skip the
1115 initial header line. The <option>--separator</option> option
1116 takes an argument which denotes what will be used between
1117 the output fields. Both these options are to help scripting.
1121 The units used to display the numeric values in the output
1122 varies, depending on the options given. By default, the values
1123 will be formatted in the most appropriate unit. If the
1124 <option>--separator</option> option is given, then the values
1125 are shown in mebibytes to allow parsing by scripts. In both
1126 cases, the <option>--units</option> option can be used to
1127 enforce a given output unit.
1131 The <option>--roman</option> option allows latin people to better
1132 understand the cluster instances' status.
1136 The <option>-o</option> option takes a comma-separated list
1137 of output fields. The available fields and their meaning
1143 <simpara>the instance name</simpara>
1149 <simpara>the OS of the instance</simpara>
1155 <simpara>the primary node of the instance</simpara>
1161 <simpara>comma-separated list of secondary nodes for the
1162 instance; usually this will be just one node</simpara>
1166 <term>admin_state</term>
1168 <simpara>the desired state of the instance (either "yes"
1169 or "no" denoting the instance should run or
1174 <term>disk_template</term>
1176 <simpara>the disk template of the instance</simpara>
1180 <term>oper_state</term>
1182 <simpara>the actual state of the instance; can be
1183 one of the values "running", "stopped", "(node
1190 <simpara>combined form of admin_state and oper_stat;
1192 <computeroutput>ERROR_nodedown</computeroutput> if the
1193 node of the instance is down,
1194 <computeroutput>ERROR_down</computeroutput> if the
1195 instance should run but is down,
1196 <computeroutput>ERROR_up</computeroutput> if the
1197 instance should be stopped but is actually running,
1198 <computeroutput>ADMIN_down</computeroutput> if the
1199 instance has been stopped (and is stopped) and
1200 <computeroutput>running</computeroutput> if the
1201 instance is set to be running (and is
1206 <term>oper_ram</term>
1208 <simpara>the actual memory usage of the instance as seen
1209 by the hypervisor</simpara>
1213 <term>oper_vcpus</term>
1215 <simpara>the actual number of VCPUs the instance is using
1216 as seen by the hypervisor</simpara>
1222 <simpara>the ip address Ganeti recognizes as associated with
1223 the first instance interface</simpara>
1229 <simpara>the first instance interface MAC address</simpara>
1234 <term>nic_mode</term>
1236 <simpara>the mode of the first instance NIC
1237 (routed or bridged)</simpara>
1241 <term>nic_link</term>
1243 <simpara>the link of the first instance NIC
1248 <term>sda_size</term>
1250 <simpara>the size of the instance's first disk</simpara>
1254 <term>sdb_size</term>
1256 <simpara>the size of the instance's second disk, if
1263 <simpara>the number of VCPUs allocated to the
1270 <simpara>comma-separated list of the instances's
1275 <term>serial_no</term>
1277 <simpara>the so called 'serial number' of the
1278 instance; this is a numeric field that is incremented
1279 each time the instance is modified, and it can be used
1280 to track modifications</simpara>
1287 the creation time of the instance; note that this
1288 field contains spaces and as such it's harder to
1292 if this attribute is not present (e.g. when
1293 upgrading from older versions), then "N/A" will be
1302 the last modification time of the instance; note
1303 that this field contains spaces and as such it's
1307 if this attribute is not present (e.g. when
1308 upgrading from older versions), then "N/A" will be
1317 <simpara>Show the UUID of the instance (generated
1318 automatically by Ganeti)</simpara>
1323 <term>network_port</term>
1325 <simpara>If the instance has a network port assigned
1326 to it (e.g. for VNC connections), this will be shown,
1327 otherwise <literal>-</literal> will be
1328 displayed.</simpara>
1332 <term>beparams</term>
1334 <simpara>A text format of the entire beparams for the
1335 instance. It's more useful to select individual fields
1336 from this dictionary, see below.</simpara>
1340 <term>disk.count</term>
1342 <simpara>The number of instance disks.</simpara>
1346 <term>disk.size/N</term>
1348 <simpara>The size of the instance's Nth disk. This is
1349 a more generic form of the <literal>sda_size</literal>
1350 and <literal>sdb_size</literal> fields.</simpara>
1354 <term>disk.sizes</term>
1356 <simpara>A comma-separated list of the disk sizes for
1357 this instance.</simpara>
1361 <term>disk_usage</term>
1363 <simpara>The total disk space used by this instance on
1364 each of its nodes. This is not the instance-visible
1365 disk size, but the actual disk "cost" of the
1370 <term>nic.mac/N</term>
1372 <simpara>The MAC of the Nth instance NIC.</simpara>
1376 <term>nic.ip/N</term>
1378 <simpara>The IP address of the Nth instance NIC.</simpara>
1382 <term>nic.mode/N</term>
1384 <simpara>The mode of the Nth instance NIC</simpara>
1388 <term>nic.link/N</term>
1390 <simpara>The link of the Nth instance NIC</simpara>
1394 <term>nic.macs</term>
1396 <simpara>A comma-separated list of all the MACs of the
1397 instance's NICs.</simpara>
1401 <term>nic.ips</term>
1403 <simpara>A comma-separated list of all the IP
1404 addresses of the instance's NICs.</simpara>
1408 <term>nic.modes</term>
1410 <simpara>A comma-separated list of all the modes of the
1411 instance's NICs.</simpara>
1415 <term>nic.links</term>
1417 <simpara>A comma-separated list of all the link parameters
1418 of the instance's NICs.</simpara>
1422 <term>nic.count</term>
1424 <simpara>The number of instance nics.</simpara>
1428 <term>hv/<replaceable>NAME</replaceable></term>
1430 <simpara>The value of the hypervisor parameter called
1431 <replaceable>NAME</replaceable>. For details of what
1432 hypervisor parameters exist and their meaning, see the
1433 <command>add</command> command.</simpara>
1437 <term>be/memory</term>
1439 <simpara>The configured memory for the instance.</simpara>
1443 <term>be/vcpus</term>
1445 <simpara>The configured number of VCPUs for the
1450 <term>be/auto_balance</term>
1452 <simpara>Whether the instance is considered in N+1
1460 If the value of the option starts with the character
1461 <constant>+</constant>, the new field(s) will be added to the
1462 default list. This allows to quickly see the default list
1463 plus a few other fields, instead of retyping the entire list
1468 There is a subtle grouping about the available output
1469 fields: all fields except for <option>oper_state</option>,
1470 <option>oper_ram</option>, <option>oper_vcpus</option> and
1471 <option>status</option> are
1472 configuration value and not run-time values. So if you don't
1473 select any of the these fields, the query will be satisfied
1474 instantly from the cluster configuration, without having to
1475 ask the remote nodes for the data. This can be helpful for
1476 big clusters when you only want some data and it makes sense
1477 to specify a reduced set of output fields.
1480 <para>The default output field list is:
1481 <simplelist type="inline">
1482 <member>name</member>
1484 <member>pnode</member>
1485 <member>admin_state</member>
1486 <member>oper_state</member>
1487 <member>oper_ram</member>
1496 <command>info</command>
1502 <group choice="req">
1504 <arg rep="repeat"><replaceable>instance</replaceable></arg>
1509 Show detailed information about the given instance(s). This is
1510 different from <command>list</command> as it shows detailed data
1511 about the instance's disks (especially useful for the drbd disk
1516 If the option <option>-s</option> is used, only information
1517 available in the configuration file is returned, without
1518 querying nodes, making the operation faster.
1522 Use the <option>--all</option> to get info about all instances,
1523 rather than explicitly passing the ones you're interested in.
1527 The <option>--roman</option> option can be used to cause envy among
1528 people who like ancient cultures, but are stuck with non-latin-friendly
1529 cluster virtualization technologies.
1535 <title>MODIFY</title>
1538 <command>modify</command>
1540 <arg choice="opt">-H <replaceable>HYPERVISOR_PARAMETERS</replaceable></arg>
1542 <arg choice="opt">-B <replaceable>BACKEND_PARAMETERS</replaceable></arg>
1545 <arg>--net add<replaceable><optional>:options</optional></replaceable></arg>
1546 <arg>--net remove</arg>
1547 <arg>--net <replaceable>N:options</replaceable></arg>
1551 <arg>--disk add:size=<replaceable>SIZE</replaceable></arg>
1552 <arg>--disk remove</arg>
1553 <arg>--disk <replaceable>N</replaceable>:mode=<replaceable>MODE</replaceable></arg>
1557 <arg>-t<group choice="req">
1563 <arg>--os-name=<replaceable>OS</replaceable> <arg>--force-variant</arg></arg>
1568 <arg choice="req"><replaceable>instance</replaceable></arg>
1572 Modifies the memory size, number of vcpus, ip address, MAC
1573 address and/or nic parameters for an instance. It can also
1574 add and remove disks and NICs to/from the instance. Note
1575 that you need to give at least one of the arguments, otherwise
1576 the command complains.
1580 The <option>-H</option> option specifies hypervisor options
1581 in the form of <userinput>name=value[,...]</userinput>. For details which options can be specified, see the <command>add</command> command.
1585 The <option>-t</option> option will change the disk template
1586 of the instance. Currently only conversions between the
1587 plain and drbd disk templates are supported, and the
1588 instance must be stopped before attempting the conversion.
1593 add:size=<replaceable>SIZE</replaceable></option> option
1594 adds a disk to the instance. The <option>--disk
1595 remove</option> will remove the last disk of the
1596 instance. The <option>--disk
1597 <replaceable>N</replaceable>:mode=<replaceable>MODE</replaceable></option>
1598 option will change the mode of the Nth disk of the instance
1599 between read-only (<literal>ro</literal>) and read-write
1600 (<literal>rw</literal>).
1605 add:<replaceable>options</replaceable></option> option will
1606 add a new NIC to the instance. The available options are the
1607 same as in the <command>add</command> command (mac, ip, link,
1608 mode). The <option>--net remove</option> will remove the
1609 last NIC of the instance, while the <option>--net
1610 <replaceable>N</replaceable>:<replaceable>options</replaceable></option>
1611 option will change the parameters of the Nth instance NIC.
1615 The option <option>--os-name</option> will change the OS
1616 name for the instance (without reinstallation). In case an
1617 OS variant is specified that is not found, then by default
1618 the modification is refused,
1619 unless <option>--force-variant</option> is passed. An
1620 invalid OS will also be refused, unless
1621 the <option>--force</option> option is given.
1625 The <option>--submit</option> option is used to send the job to
1626 the master daemon but not wait for its completion. The job
1627 ID will be shown so that it can be examined via
1628 <command>gnt-job info</command>.
1632 All the changes take effect at the next restart. If the
1633 instance is running, there is no effect on the instance.
1638 <title>REINSTALL</title>
1641 <command>reinstall</command>
1642 <arg choice="opt">-o <replaceable>os-type</replaceable></arg>
1643 <arg>--select-os</arg>
1644 <arg choice="opt">-f <replaceable>force</replaceable></arg>
1645 <arg>--force-multiple</arg>
1647 <group choice="opt">
1648 <arg>--instance</arg>
1650 <arg>--primary</arg>
1651 <arg>--secondary</arg>
1655 <arg choice="opt">-O <replaceable>OS_PARAMETERS</replaceable></arg>
1657 <arg choice="opt" rep="repeat"><replaceable>instance</replaceable></arg>
1661 Reinstalls the operating system on the given instance(s). The
1662 instance(s) must be stopped when running this command. If the
1663 <option>--os-type</option> is specified, the operating
1668 The <option>--select-os</option> option switches to an
1669 interactive OS reinstall. The user is prompted to select the OS
1670 template from the list of available OS templates. OS parameters
1671 can be overridden using <option>-O</option>.
1675 Since this is a potentially dangerous command, the user will
1676 be required to confirm this action, unless the
1677 <option>-f</option> flag is passed. When multiple instances
1678 are selected (either by passing multiple arguments or by
1679 using the <option>--node</option>,
1680 <option>--primary</option>, <option>--secondary</option> or
1681 <option>--all</option> options), the user must pass the
1682 <option>--force-multiple</option> options to skip the
1683 interactive confirmation.
1687 The <option>--submit</option> option is used to send the job to
1688 the master daemon but not wait for its completion. The job
1689 ID will be shown so that it can be examined via
1690 <command>gnt-job info</command>.
1696 <title>RENAME</title>
1699 <command>rename</command>
1700 <arg>--no-ip-check</arg>
1701 <arg>--no-name-check</arg>
1703 <arg choice="req"><replaceable>instance</replaceable></arg>
1704 <arg choice="req"><replaceable>new_name</replaceable></arg>
1708 Renames the given instance. The instance must be stopped
1709 when running this command. The requirements for the new name
1710 are the same as for adding an instance: the new name must be
1711 resolvable and the IP it resolves to must not be reachable
1712 (in order to prevent duplicate IPs the next time the
1713 instance is started). The IP test can be skipped if the
1714 <option>--no-ip-check</option> option is passed.
1718 The <option>--no-name-check</option> skips the check for the
1719 new instance name via the resolver (e.g. in DNS or /etc/hosts,
1720 depending on your setup). Since the name check is used to
1721 compute the IP address, if you pass this option you must
1722 also pass the <option>--no-ip-check</option> option.
1726 The <option>--submit</option> option is used to send the job to
1727 the master daemon but not wait for its completion. The job
1728 ID will be shown so that it can be examined via
1729 <command>gnt-job info</command>.
1737 <title>Starting/stopping/connecting to console</title>
1740 <title>STARTUP</title>
1743 <command>startup</command>
1746 <arg>--ignore-offline</arg>
1748 <arg>--force-multiple</arg>
1750 <group choice="opt">
1751 <arg>--instance</arg>
1753 <arg>--primary</arg>
1754 <arg>--secondary</arg>
1757 <arg>--node-tags</arg>
1758 <arg>--pri-node-tags</arg>
1759 <arg>--sec-node-tags</arg>
1762 <arg>-H <option>key=value...</option></arg>
1763 <arg>-B <option>key=value...</option></arg>
1768 rep="repeat"><replaceable>name</replaceable></arg>
1772 Starts one or more instances, depending on the following
1773 options. The four available modes are:
1776 <term><option>--instance</option></term>
1778 <simpara>will start the instances given as arguments
1779 (at least one argument required); this is the default
1786 <simpara>will start the instances who have the given
1787 node as either primary or secondary</simpara>
1791 <term><option>--primary</option></term>
1793 <simpara>will start all instances whose primary node
1794 is in the list of nodes passed as arguments (at least
1795 one node required)</simpara>
1799 <term><option>--secondary</option></term>
1801 <simpara>will start all instances whose secondary node
1802 is in the list of nodes passed as arguments (at least
1803 one node required)</simpara>
1809 <simpara>will start all instances in the cluster (no
1810 arguments accepted)</simpara>
1816 <simpara>will start all instances in the cluster with
1817 the tags given as arguments</simpara>
1821 <term>--node-tags</term>
1823 <simpara>will start all instances in the cluster on
1824 nodes with the tags given as arguments</simpara>
1828 <term>--pri-node-tags</term>
1830 <simpara>will start all instances in the cluster on
1831 primary nodes with the tags given as
1836 <term>--sec-node-tags</term>
1838 <simpara>will start all instances in the cluster on
1839 secondary nodes with the tags given as
1847 Note that although you can pass more than one selection
1848 option, the last one wins, so in order to guarantee the
1849 desired result, don't pass more than one such option.
1853 Use <option>--force</option> to start even if secondary disks are
1854 failing. <option>--ignore-offline</option> can be used to ignore
1855 offline primary nodes and mark the instance as started even if
1856 the primary is not available.
1860 The <option>--force-multiple</option> will skip the
1861 interactive confirmation in the case the more than one
1862 instance will be affected.
1866 The <option>-H</option> and <option>-B</option> options
1867 specify temporary hypervisor and backend parameters that can
1868 be used to start an instance with modified parameters. They
1869 can be useful for quick testing without having to modify an
1870 instance back and forth, e.g.:
1872 # gnt-instance start -H root_args="single" instance1
1873 # gnt-instance start -B memory=2048 instance2
1875 The first form will start the instance
1876 <userinput>instance1</userinput> in single-user mode, and
1877 the instance <userinput>instance2</userinput> with 2GB of
1878 RAM (this time only, unless that is the actual instance
1879 memory size already). Note that the values override the
1880 instance parameters (and not extend them): an instance with
1881 "root_args=ro" when started with <userinput>-H
1882 root_args=single</userinput> will result in "single", not
1887 The <option>--submit</option> option is used to send the job to
1888 the master daemon but not wait for its completion. The job
1889 ID will be shown so that it can be examined via
1890 <command>gnt-job info</command>.
1896 # gnt-instance start instance1.example.com
1897 # gnt-instance start --node node1.example.com node2.example.com
1898 # gnt-instance start --all
1904 <title>SHUTDOWN</title>
1907 <command>shutdown</command>
1909 <arg>--timeout=<replaceable>N</replaceable></arg>
1911 <arg>--force-multiple</arg>
1912 <arg>--ignore-offline</arg>
1914 <group choice="opt">
1915 <arg>--instance</arg>
1917 <arg>--primary</arg>
1918 <arg>--secondary</arg>
1921 <arg>--node-tags</arg>
1922 <arg>--pri-node-tags</arg>
1923 <arg>--sec-node-tags</arg>
1929 rep="repeat"><replaceable>name</replaceable></arg>
1933 Stops one or more instances. If the instance cannot be
1934 cleanly stopped during a hardcoded interval (currently 2
1935 minutes), it will forcibly stop the instance (equivalent to
1936 switching off the power on a physical machine).
1940 The <option>--timeout</option> is used to specify how much time to
1941 wait before forcing the shutdown (xm destroy in xen, killing the kvm
1942 process, for kvm). By default two minutes are given to each instance
1947 The <option>--instance</option>, <option>--node</option>,
1948 <option>--primary</option>, <option>--secondary</option>,
1949 <option>--all</option>, <option>--tags</option>,
1950 <option>--node-tags</option>, <option>--pri-node-tags</option> and
1951 <option>--sec-node-tags</option> options are similar as for the
1952 <command>startup</command> command and they influence the
1953 actual instances being shutdown.
1957 The <option>--submit</option> option is used to send the job to
1958 the master daemon but not wait for its completion. The job
1959 ID will be shown so that it can be examined via
1960 <command>gnt-job info</command>.
1964 <option>--ignore-offline</option> can be used to ignore offline
1965 primary nodes and force the instance to be marked as stopped. This
1966 option should be used with care as it can lead to an
1967 inconsistent cluster state.
1973 # gnt-instance shutdown instance1.example.com
1974 # gnt-instance shutdown --all
1980 <title>REBOOT</title>
1983 <command>reboot</command>
1985 <arg>--type=<replaceable>REBOOT-TYPE</replaceable></arg>
1987 <arg>--ignore-secondaries</arg>
1989 <arg>--shutdown-timeout=<replaceable>N</replaceable></arg>
1991 <arg>--force-multiple</arg>
1993 <group choice="opt">
1994 <arg>--instance</arg>
1996 <arg>--primary</arg>
1997 <arg>--secondary</arg>
2000 <arg>--node-tags</arg>
2001 <arg>--pri-node-tags</arg>
2002 <arg>--sec-node-tags</arg>
2008 rep="repeat"><replaceable>name</replaceable></arg>
2012 Reboots one or more instances. The type of reboot depends on
2013 the value of <option>--type</option>. A soft reboot does a
2014 hypervisor reboot, a hard reboot does a instance stop,
2015 recreates the hypervisor config for the instance and
2016 starts the instance. A full reboot does the equivalent
2017 of <command>gnt-instance shutdown && gnt-instance
2018 startup</command>. The default is hard reboot.
2022 For the hard reboot the option
2023 <option>--ignore-secondaries</option> ignores errors for the
2024 secondary node while re-assembling the instance disks.
2028 The <option>--instance</option>, <option>--node</option>,
2029 <option>--primary</option>, <option>--secondary</option>,
2030 <option>--all</option>, <option>--tags</option>,
2031 <option>--node-tags</option>, <option>--pri-node-tags</option> and
2032 <option>--sec-node-tags</option> options are similar as for the
2033 <command>startup</command> command and they influence the
2034 actual instances being rebooted.
2038 The <option>--shutdown-timeout</option> is used to specify how
2039 much time to wait before forcing the shutdown (xm destroy in xen,
2040 killing the kvm process, for kvm). By default two minutes are
2041 given to each instance to stop.
2045 The <option>--force-multiple</option> will skip the
2046 interactive confirmation in the case the more than one
2047 instance will be affected.
2053 # gnt-instance reboot instance1.example.com
2054 # gnt-instance reboot --type=full instance1.example.com
2060 <title>CONSOLE</title>
2062 <command>console</command>
2063 <arg choice="opt">--show-cmd</arg>
2064 <arg choice="req"><replaceable>instance</replaceable></arg>
2068 Connects to the console of the given instance. If the
2069 instance is not up, an error is returned. Use the
2070 <option>--show-cmd</option> option to display the command
2071 instead of executing it.
2075 For HVM instances, this will attempt to connect to the
2076 serial console of the instance. To connect to the
2077 virtualized "physical" console of a HVM instance, use a VNC
2078 client with the connection info from the
2079 <command>info</command> command.
2085 # gnt-instance console instance1.example.com
2093 <title>Disk management</title>
2096 <title>REPLACE-DISKS</title>
2099 <command>replace-disks</command>
2101 <arg>--early-release</arg>
2102 <arg choice="req">-p</arg>
2103 <arg>--disks <replaceable>idx</replaceable></arg>
2104 <arg choice="req"><replaceable>instance</replaceable></arg>
2108 <command>replace-disks</command>
2110 <arg>--early-release</arg>
2111 <arg choice="req">-s</arg>
2112 <arg>--disks <replaceable>idx</replaceable></arg>
2113 <arg choice="req"><replaceable>instance</replaceable></arg>
2117 <command>replace-disks</command>
2119 <arg>--early-release</arg>
2120 <group choice="req">
2121 <arg>--iallocator <replaceable>name</replaceable></arg>
2122 <arg>--new-secondary <replaceable>NODE</replaceable></arg>
2125 <arg choice="req"><replaceable>instance</replaceable></arg>
2129 <command>replace-disks</command>
2131 <arg>--early-release</arg>
2132 <arg choice="req">--auto</arg>
2133 <arg choice="req"><replaceable>instance</replaceable></arg>
2137 This command is a generalized form for replacing disks. It
2138 is currently only valid for the mirrored (DRBD) disk
2143 The first form (when passing the <option>-p</option> option)
2144 will replace the disks on the primary, while the second form
2145 (when passing the <option>-s</option> option will replace
2146 the disks on the secondary node. For these two cases (as the
2147 node doesn't change), it is possible to only run the replace
2148 for a subset of the disks, using the option
2149 <option>--disks</option> which takes a list of
2150 comma-delimited disk indices (zero-based),
2151 e.g. <userinput>0,2</userinput> to replace only the first
2156 The third form (when passing either the
2157 <option>--iallocator</option> or the
2158 <option>--new-secondary</option> option) is designed to
2159 change secondary node of the instance. Specifying
2160 <option>--iallocator</option> makes the new secondary be
2161 selected automatically by the specified allocator plugin,
2162 otherwise the new secondary node will be the one chosen
2163 manually via the <option>--new-secondary</option> option.
2167 The fourth form (when using <option>--auto</option>) will
2168 automatically determine which disks of an instance are faulty and
2169 replace them within the same node. The <option>--auto</option>
2170 option works only when an instance has only faulty disks on
2171 either the primary or secondary node; it doesn't work when
2172 both sides have faulty disks.
2176 The <option>--submit</option> option is used to send the job to
2177 the master daemon but not wait for its completion. The job
2178 ID will be shown so that it can be examined via
2179 <command>gnt-job info</command>.
2183 The <option>--early-release</option> changes the code so
2184 that the old storage on secondary node(s) is removed early
2185 (before the resync is completed) and the internal Ganeti
2186 locks for the current (and new, if any) secondary node are
2187 also released, thus allowing more parallelism in the cluster
2188 operation. This should be used only when recovering from a
2189 disk failure on the current secondary (thus the old storage
2190 is already broken) or when the storage on the primary node
2191 is known to be fine (thus we won't need the old storage for
2192 potential recovery).
2196 Note that it is not possible to select an offline or drained
2197 node as a new secondary.
2203 <title>ACTIVATE-DISKS</title>
2206 <command>activate-disks</command>
2208 <arg>--ignore-size</arg>
2209 <arg choice="req"><replaceable>instance</replaceable></arg>
2212 Activates the block devices of the given instance. If
2213 successful, the command will show the location and name of
2216 node1.example.com:disk/0:/dev/drbd0
2217 node1.example.com:disk/1:/dev/drbd1
2220 In this example, <emphasis>node1.example.com</emphasis> is
2221 the name of the node on which the devices have been
2222 activated. The <emphasis>disk/0</emphasis> and
2223 <emphasis>disk/1</emphasis> are the Ganeti-names of the
2224 instance disks; how they are visible inside the instance is
2225 hypervisor-specific. <emphasis>/dev/drbd0</emphasis> and
2226 <emphasis>/dev/drbd1</emphasis> are the actual block devices
2227 as visible on the node.
2231 The <option>--submit</option> option is used to send the job to
2232 the master daemon but not wait for its completion. The job
2233 ID will be shown so that it can be examined via
2234 <command>gnt-job info</command>.
2238 The <option>--ignore-size</option> option can be used to
2239 activate disks ignoring the currently configured size in
2240 Ganeti. This can be used in cases where the configuration
2241 has gotten out of sync with the real-world (e.g. after a
2242 partially-failed grow-disk operation or due to rounding in
2243 LVM devices). This should not be used in normal cases, but
2244 only when activate-disks fails without it.
2248 Note that it is safe to run this command while the instance
2254 <title>DEACTIVATE-DISKS</title>
2257 <command>deactivate-disks</command>
2259 <arg choice="req"><replaceable>instance</replaceable></arg>
2262 De-activates the block devices of the given instance. Note
2263 that if you run this command for an instance with a drbd
2264 disk template, while it is running, it will not be able to
2265 shutdown the block devices on the primary node, but it will
2266 shutdown the block devices on the secondary nodes, thus
2267 breaking the replication.
2271 The <option>--submit</option> option is used to send the job to
2272 the master daemon but not wait for its completion. The job
2273 ID will be shown so that it can be examined via
2274 <command>gnt-job info</command>.
2280 <title>GROW-DISK</title>
2282 <command>grow-disk</command>
2283 <arg>--no-wait-for-sync</arg>
2285 <arg choice="req"><replaceable>instance</replaceable></arg>
2286 <arg choice="req"><replaceable>disk</replaceable></arg>
2287 <arg choice="req"><replaceable>amount</replaceable></arg>
2291 Grows an instance's disk. This is only possible for
2292 instances having a <literal>plain</literal> or
2293 <literal>drbd</literal> disk template.
2297 Note that this command only change the block device size; it
2298 will not grow the actual filesystems, partitions, etc. that
2299 live on that disk. Usually, you will need to:
2302 <simpara>use <command>gnt-instance grow-disk</command></simpara>
2305 <simpara>reboot the instance (later, at a convenient
2309 <simpara>use a filesystem resizer, such as
2310 <citerefentry> <refentrytitle>ext2online</refentrytitle>
2311 <manvolnum>8</manvolnum> </citerefentry> or
2312 <citerefentry> <refentrytitle>xfs_growfs</refentrytitle>
2313 <manvolnum>8</manvolnum> </citerefentry> to resize the
2314 filesystem, or use <citerefentry>
2315 <refentrytitle>fdisk</refentrytitle>
2316 <manvolnum>8</manvolnum> </citerefentry> to change the
2317 partition table on the disk
2325 The <replaceable>disk</replaceable> argument is the index of
2326 the instance disk to grow. The
2327 <replaceable>amount</replaceable> argument is given either
2328 as a number (and it represents the amount to increase the
2329 disk with in mebibytes) or can be given similar to the
2330 arguments in the create instance operation, with a suffix
2335 Note that the disk grow operation might complete on one node
2336 but fail on the other; this will leave the instance with
2337 different-sized LVs on the two nodes, but this will not
2338 create problems (except for unused space).
2342 If you do not want gnt-instance to wait for the new disk
2343 region to be synced, use the
2344 <option>--no-wait-for-sync</option> option.
2348 The <option>--submit</option> option is used to send the job to
2349 the master daemon but not wait for its completion. The job
2350 ID will be shown so that it can be examined via
2351 <command>gnt-job info</command>.
2355 <para>Example (increase the first disk for instance1 by 16GiB):
2357 # gnt-instance grow-disk instance1.example.com 0 16g
2362 Also note that disk shrinking is not supported; use
2363 <command>gnt-backup export</command> and then
2364 <command>gnt-backup import</command> to reduce the disk size
2370 <title>RECREATE-DISKS</title>
2373 <command>recreate-disks</command>
2375 <arg>--disks=<option>indices</option></arg>
2376 <arg choice="req"><replaceable>instance</replaceable></arg>
2379 Recreates the disks of the given instance, or only a subset
2380 of the disks (if the option <option>disks</option> is
2381 passed, which must be a comma-separated list of disk
2382 indices, starting from zero).
2386 Note that this functionality should only be used for missing
2387 disks; if any of the given disks already exists, the
2388 operation will fail. While this is suboptimal,
2389 recreate-disks should hopefully not be needed in normal
2390 operation and as such the impact of this is low.
2394 The <option>--submit</option> option is used to send the job to
2395 the master daemon but not wait for its completion. The job
2396 ID will be shown so that it can be examined via
2397 <command>gnt-job info</command>.
2405 <title>Recovery</title>
2408 <title>FAILOVER</title>
2411 <command>failover</command>
2413 <arg>--ignore-consistency</arg>
2414 <arg>--shutdown-timeout=<replaceable>N</replaceable></arg>
2416 <arg choice="req"><replaceable>instance</replaceable></arg>
2420 Failover will fail the instance over its secondary
2421 node. This works only for instances having a drbd disk
2426 Normally the failover will check the consistency of the
2427 disks before failing over the instance. If you are trying to
2428 migrate instances off a dead node, this will fail. Use the
2429 <option>--ignore-consistency</option> option for this
2430 purpose. Note that this option can be dangerous as errors in
2431 shutting down the instance will be ignored, resulting in
2432 possibly having the instance running on two machines in
2433 parallel (on disconnected DRBD drives).
2437 The <option>--shutdown-timeout</option> is used to specify how
2438 much time to wait before forcing the shutdown (xm destroy in xen,
2439 killing the kvm process, for kvm). By default two minutes are
2440 given to each instance to stop.
2444 The <option>--submit</option> option is used to send the job to
2445 the master daemon but not wait for its completion. The job
2446 ID will be shown so that it can be examined via
2447 <command>gnt-job info</command>.
2453 # gnt-instance failover instance1.example.com
2459 <title>MIGRATE</title>
2462 <command>migrate</command>
2464 <arg choice="req">--cleanup</arg>
2465 <arg choice="req"><replaceable>instance</replaceable></arg>
2469 <command>migrate</command>
2471 <arg>--non-live</arg>
2472 <arg>--migration-mode=live|non-live</arg>
2473 <arg choice="req"><replaceable>instance</replaceable></arg>
2477 Migrate will move the instance to its secondary node without
2478 shutdown. It only works for instances having the drbd8 disk
2483 The migration command needs a perfectly healthy instance, as
2484 we rely on the dual-master capability of drbd8 and the disks
2485 of the instance are not allowed to be degraded.
2489 The <option>--non-live</option>
2490 and <option>--migration-mode=non-live</option> options will
2491 switch (for the hypervisors that support it) between a
2492 "fully live" (i.e. the interruption is as minimal as
2493 possible) migration and one in which the instance is frozen,
2494 its state saved and transported to the remote node, and then
2495 resumed there. This all depends on the hypervisor support
2496 for two different methods. In any case, it is not an error
2497 to pass this parameter (it will just be ignored if the
2498 hypervisor doesn't support it). The
2499 option <option>--migration-mode=live</option> option will
2500 request a fully-live migration. The default, when neither
2501 option is passed, depends on the hypervisor parameters (and
2502 can be viewed with the <command>gnt-cluster info</command>
2507 If the <option>--cleanup</option> option is passed, the
2508 operation changes from migration to attempting recovery from
2509 a failed previous migration. In this mode, Ganeti checks if
2510 the instance runs on the correct node (and updates its
2511 configuration if not) and ensures the instances's disks are
2512 configured correctly. In this mode, the
2513 <option>--non-live</option> option is ignored.
2517 The option <option>-f</option> will skip the prompting for
2521 Example (and expected output):
2523 # gnt-instance migrate instance1
2524 Migrate will happen to the instance instance1. Note that migration is
2525 **experimental** in this version. This might impact the instance if
2526 anything goes wrong. Continue?
2528 * checking disk consistency between source and target
2529 * ensuring the target is in secondary mode
2530 * changing disks into dual-master mode
2531 - INFO: Waiting for instance instance1 to sync disks.
2532 - INFO: Instance instance1's disks are in sync.
2533 * migrating instance to node2.example.com
2534 * changing the instance's disks on source node to secondary
2535 - INFO: Waiting for instance instance1 to sync disks.
2536 - INFO: Instance instance1's disks are in sync.
2537 * changing the instance's disks to single-master
2547 <command>move</command>
2549 <arg>-n <replaceable>node</replaceable></arg>
2550 <arg>--shutdown-timeout=<replaceable>N</replaceable></arg>
2552 <arg choice="req"><replaceable>instance</replaceable></arg>
2556 Move will move the instance to an arbitrary node in the
2557 cluster. This works only for instances having a plain or
2562 Note that since this operation is done via data copy, it
2563 will take a long time for big disks (similar to
2564 replace-disks for a drbd instance).
2568 The <option>--shutdown-timeout</option> is used to specify how
2569 much time to wait before forcing the shutdown (xm destroy in xen,
2570 killing the kvm process, for kvm). By default two minutes are
2571 given to each instance to stop.
2575 The <option>--submit</option> option is used to send the job to
2576 the master daemon but not wait for its completion. The job
2577 ID will be shown so that it can be examined via
2578 <command>gnt-job info</command>.
2584 # gnt-instance move -n node3.example.com instance1.example.com
2595 <title>ADD-TAGS</title>
2598 <command>add-tags</command>
2599 <arg choice="opt">--from <replaceable>file</replaceable></arg>
2600 <arg choice="req"><replaceable>instancename</replaceable></arg>
2602 rep="repeat"><replaceable>tag</replaceable></arg>
2606 Add tags to the given instance. If any of the tags contains
2607 invalid characters, the entire operation will abort.
2610 If the <option>--from</option> option is given, the list of
2611 tags will be extended with the contents of that file (each
2612 line becomes a tag). In this case, there is not need to pass
2613 tags on the command line (if you do, both sources will be
2614 used). A file name of - will be interpreted as stdin.
2619 <title>LIST-TAGS</title>
2622 <command>list-tags</command>
2623 <arg choice="req"><replaceable>instancename</replaceable></arg>
2626 <para>List the tags of the given instance.</para>
2630 <title>REMOVE-TAGS</title>
2632 <command>remove-tags</command>
2633 <arg choice="opt">--from <replaceable>file</replaceable></arg>
2634 <arg choice="req"><replaceable>instancename</replaceable></arg>
2636 rep="repeat"><replaceable>tag</replaceable></arg>
2640 Remove tags from the given instance. If any of the tags are
2641 not existing on the node, the entire operation will abort.
2645 If the <option>--from</option> option is given, the list of
2646 tags will be extended with the contents of that file (each
2647 line becomes a tag). In this case, there is not need to pass
2648 tags on the command line (if you do, both sources will be
2649 used). A file name of - will be interpreted as stdin.
2661 <!-- Keep this comment at the end of the file
2666 sgml-minimize-attributes:nil
2667 sgml-always-quote-attributes:t
2670 sgml-parent-document:nil
2671 sgml-default-dtd-file:nil
2672 sgml-exposed-tags:nil
2673 sgml-local-catalogs:nil
2674 sgml-local-ecat-files:nil