10 pvecm - Proxmox VE Cluster Manager
15 include::pvecm.1-synopsis.adoc[]
27 The {PVE} cluster manager `pvecm` is a tool to create a group of
28 physical servers. Such a group is called a *cluster*. We use the
29 http://www.corosync.org[Corosync Cluster Engine] for reliable group
30 communication, and such clusters can consist of up to 32 physical nodes
31 (probably more, dependent on network latency).
33 `pvecm` can be used to create a new cluster, join nodes to a cluster,
34 leave the cluster, get status information and do various other cluster
35 related tasks. The **P**rox**m**o**x** **C**luster **F**ile **S**ystem (``pmxcfs'')
36 is used to transparently distribute the cluster configuration to all cluster
39 Grouping nodes into a cluster has the following advantages:
41 * Centralized, web based management
43 * Multi-master clusters: each node can do all management task
45 * `pmxcfs`: database-driven file system for storing configuration files,
46 replicated in real-time on all nodes using `corosync`.
48 * Easy migration of virtual machines and containers between physical
53 * Cluster-wide services like firewall and HA
59 * All nodes must be in the same network as `corosync` uses IP Multicast
60 to communicate between nodes (also see
61 http://www.corosync.org[Corosync Cluster Engine]). Corosync uses UDP
62 ports 5404 and 5405 for cluster communication.
64 NOTE: Some switches do not support IP multicast by default and must be
65 manually enabled first.
67 * Date and time have to be synchronized.
69 * SSH tunnel on TCP port 22 between nodes is used.
71 * If you are interested in High Availability, you need to have at
72 least three nodes for reliable quorum. All nodes should have the
75 * We recommend a dedicated NIC for the cluster traffic, especially if
76 you use shared storage.
78 NOTE: It is not possible to mix Proxmox VE 3.x and earlier with
79 Proxmox VE 4.0 cluster nodes.
85 First, install {PVE} on all nodes. Make sure that each node is
86 installed with the final hostname and IP configuration. Changing the
87 hostname and IP is not possible after cluster creation.
89 Currently the cluster creation has to be done on the console, so you
90 need to login via `ssh`.
95 Login via `ssh` to the first {pve} node. Use a unique name for your cluster.
96 This name cannot be changed later.
98 hp1# pvecm create YOUR-CLUSTER-NAME
100 CAUTION: The cluster name is used to compute the default multicast
101 address. Please use unique cluster names if you run more than one
102 cluster inside your network.
104 To check the state of your cluster use:
108 Multiple Clusters In Same Network
109 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
111 It is possible to create multiple clusters in the same physical or logical
112 network. Each cluster must have a unique name, which is used to generate the
113 cluster's multicast group address. As long as no duplicate cluster names are
114 configured in one network segment, the different clusters won't interfere with
117 If multiple clusters operate in a single network it may be beneficial to setup
118 an IGMP querier and enable IGMP Snooping in said network. This may reduce the
119 load of the network significantly because multicast packets are only delivered
120 to endpoints of the respective member nodes.
123 Adding Nodes to the Cluster
124 ---------------------------
126 Login via `ssh` to the node you want to add.
128 hp2# pvecm add IP-ADDRESS-CLUSTER
130 For `IP-ADDRESS-CLUSTER` use the IP from an existing cluster node.
132 CAUTION: A new node cannot hold any VMs, because you would get
133 conflicts about identical VM IDs. Also, all existing configuration in
134 `/etc/pve` is overwritten when you join a new node to the cluster. To
135 workaround, use `vzdump` to backup and restore to a different VMID after
136 adding the node to the cluster.
138 To check the state of cluster:
142 .Cluster status after adding 4 nodes
147 Date: Mon Apr 20 12:30:13 2015
148 Quorum provider: corosync_votequorum
154 Votequorum information
155 ~~~~~~~~~~~~~~~~~~~~~~
162 Membership information
163 ~~~~~~~~~~~~~~~~~~~~~~
165 0x00000001 1 192.168.15.91
166 0x00000002 1 192.168.15.92 (local)
167 0x00000003 1 192.168.15.93
168 0x00000004 1 192.168.15.94
171 If you only want the list of all nodes use:
175 .List nodes in a cluster
179 Membership information
180 ~~~~~~~~~~~~~~~~~~~~~~
188 [[adding-nodes-with-separated-cluster-network]]
189 Adding Nodes With Separated Cluster Network
190 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
192 When adding a node to a cluster with a separated cluster network you need to
193 use the 'ringX_addr' parameters to set the nodes address on those networks:
197 pvecm add IP-ADDRESS-CLUSTER -ring0_addr IP-ADDRESS-RING0
200 If you want to use the Redundant Ring Protocol you will also want to pass the
201 'ring1_addr' parameter.
204 Remove a Cluster Node
205 ---------------------
207 CAUTION: Read carefully the procedure before proceeding, as it could
208 not be what you want or need.
210 Move all virtual machines from the node. Make sure you have no local
211 data or backups you want to keep, or save them accordingly.
212 In the following example we will remove the node hp4 from the cluster.
214 Log in to a *different* cluster node (not hp4), and issue a `pvecm nodes`
215 command to identify the node ID to remove:
220 Membership information
221 ~~~~~~~~~~~~~~~~~~~~~~
230 At this point you must power off hp4 and
231 make sure that it will not power on again (in the network) as it
234 IMPORTANT: As said above, it is critical to power off the node
235 *before* removal, and make sure that it will *never* power on again
236 (in the existing cluster network) as it is.
237 If you power on the node as it is, your cluster will be screwed up and
238 it could be difficult to restore a clean cluster state.
240 After powering off the node hp4, we can safely remove it from the cluster.
242 hp1# pvecm delnode hp4
244 If the operation succeeds no output is returned, just check the node
245 list again with `pvecm nodes` or `pvecm status`. You should see
253 Date: Mon Apr 20 12:44:28 2015
254 Quorum provider: corosync_votequorum
260 Votequorum information
261 ~~~~~~~~~~~~~~~~~~~~~~
268 Membership information
269 ~~~~~~~~~~~~~~~~~~~~~~
271 0x00000001 1 192.168.15.90 (local)
272 0x00000002 1 192.168.15.91
273 0x00000003 1 192.168.15.92
276 If, for whatever reason, you want that this server joins the same
277 cluster again, you have to
279 * reinstall {pve} on it from scratch
281 * then join it, as explained in the previous section.
283 [[pvecm_separate_node_without_reinstall]]
284 Separate A Node Without Reinstalling
285 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
287 CAUTION: This is *not* the recommended method, proceed with caution. Use the
288 above mentioned method if you're unsure.
290 You can also separate a node from a cluster without reinstalling it from
291 scratch. But after removing the node from the cluster it will still have
292 access to the shared storages! This must be resolved before you start removing
293 the node from the cluster. A {pve} cluster cannot share the exact same
294 storage with another cluster, as storage locking doesn't work over cluster
295 boundary. Further, it may also lead to VMID conflicts.
297 Its suggested that you create a new storage where only the node which you want
298 to separate has access. This can be an new export on your NFS or a new Ceph
299 pool, to name a few examples. Its just important that the exact same storage
300 does not gets accessed by multiple clusters. After setting this storage up move
301 all data from the node and its VMs to it. Then you are ready to separate the
302 node from the cluster.
304 WARNING: Ensure all shared resources are cleanly separated! You will run into
305 conflicts and problems else.
307 First stop the corosync and the pve-cluster services on the node:
310 systemctl stop pve-cluster
311 systemctl stop corosync
314 Start the cluster filesystem again in local mode:
320 Delete the corosync configuration files:
323 rm /etc/pve/corosync.conf
327 You can now start the filesystem again as normal service:
331 systemctl start pve-cluster
334 The node is now separated from the cluster. You can deleted it from a remaining
335 node of the cluster with:
338 pvecm delnode oldnode
341 If the command failed, because the remaining node in the cluster lost quorum
342 when the now separate node exited, you may set the expected votes to 1 as a workaround:
348 And the repeat the 'pvecm delnode' command.
350 Now switch back to the separated node, here delete all remaining files left
351 from the old cluster. This ensures that the node can be added to another
352 cluster again without problems.
356 rm /var/lib/corosync/*
359 As the configuration files from the other nodes are still in the cluster
360 filesystem you may want to clean those up too. Remove simply the whole
361 directory recursive from '/etc/pve/nodes/NODENAME', but check three times that
362 you used the correct one before deleting it.
364 CAUTION: The nodes SSH keys are still in the 'authorized_key' file, this means
365 the nodes can still connect to each other with public key authentication. This
366 should be fixed by removing the respective keys from the
367 '/etc/pve/priv/authorized_keys' file.
372 {pve} use a quorum-based technique to provide a consistent state among
375 [quote, from Wikipedia, Quorum (distributed computing)]
377 A quorum is the minimum number of votes that a distributed transaction
378 has to obtain in order to be allowed to perform an operation in a
382 In case of network partitioning, state changes requires that a
383 majority of nodes are online. The cluster switches to read-only mode
386 NOTE: {pve} assigns a single vote to each node by default.
391 The cluster network is the core of a cluster. All messages sent over it have to
392 be delivered reliable to all nodes in their respective order. In {pve} this
393 part is done by corosync, an implementation of a high performance low overhead
394 high availability development toolkit. It serves our decentralized
395 configuration file system (`pmxcfs`).
397 [[cluster-network-requirements]]
400 This needs a reliable network with latencies under 2 milliseconds (LAN
401 performance) to work properly. While corosync can also use unicast for
402 communication between nodes its **highly recommended** to have a multicast
403 capable network. The network should not be used heavily by other members,
404 ideally corosync runs on its own network.
405 *never* share it with network where storage communicates too.
407 Before setting up a cluster it is good practice to check if the network is fit
410 * Ensure that all nodes are in the same subnet. This must only be true for the
411 network interfaces used for cluster communication (corosync).
413 * Ensure all nodes can reach each other over those interfaces, using `ping` is
414 enough for a basic test.
416 * Ensure that multicast works in general and a high package rates. This can be
417 done with the `omping` tool. The final "%loss" number should be < 1%.
421 omping -c 10000 -i 0.001 -F -q NODE1-IP NODE2-IP ...
424 * Ensure that multicast communication works over an extended period of time.
425 This uncovers problems where IGMP snooping is activated on the network but
426 no multicast querier is active. This test has a duration of around 10
431 omping -c 600 -i 1 -q NODE1-IP NODE2-IP ...
434 Your network is not ready for clustering if any of these test fails. Recheck
435 your network configuration. Especially switches are notorious for having
436 multicast disabled by default or IGMP snooping enabled with no IGMP querier
439 In smaller cluster its also an option to use unicast if you really cannot get
442 Separate Cluster Network
443 ~~~~~~~~~~~~~~~~~~~~~~~~
445 When creating a cluster without any parameters the cluster network is generally
446 shared with the Web UI and the VMs and its traffic. Depending on your setup
447 even storage traffic may get sent over the same network. Its recommended to
448 change that, as corosync is a time critical real time application.
450 Setting Up A New Network
451 ^^^^^^^^^^^^^^^^^^^^^^^^
453 First you have to setup a new network interface. It should be on a physical
454 separate network. Ensure that your network fulfills the
455 <<cluster-network-requirements,cluster network requirements>>.
457 Separate On Cluster Creation
458 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
460 This is possible through the 'ring0_addr' and 'bindnet0_addr' parameter of
461 the 'pvecm create' command used for creating a new cluster.
463 If you have setup an additional NIC with a static address on 10.10.10.1/25
464 and want to send and receive all cluster communication over this interface
469 pvecm create test --ring0_addr 10.10.10.1 --bindnet0_addr 10.10.10.0
472 To check if everything is working properly execute:
475 systemctl status corosync
478 Afterwards, proceed as descripted in the section to
479 <<adding-nodes-with-separated-cluster-network,add nodes with a separated cluster network>>.
481 [[separate-cluster-net-after-creation]]
482 Separate After Cluster Creation
483 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
485 You can do this also if you have already created a cluster and want to switch
486 its communication to another network, without rebuilding the whole cluster.
487 This change may lead to short durations of quorum loss in the cluster, as nodes
488 have to restart corosync and come up one after the other on the new network.
490 Check how to <<edit-corosync-conf,edit the corosync.conf file>> first.
491 The open it and you should see a file similar to:
525 provider: corosync_votequorum
529 cluster_name: thomas-testcluster
535 bindnetaddr: 192.168.30.50
542 The first you want to do is add the 'name' properties in the node entries if
543 you do not see them already. Those *must* match the node name.
545 Then replace the address from the 'ring0_addr' properties with the new
546 addresses. You may use plain IP addresses or also hostnames here. If you use
547 hostnames ensure that they are resolvable from all nodes.
549 In my example I want to switch my cluster communication to the 10.10.10.1/25
550 network. So I replace all 'ring0_addr' respectively. I also set the bindnetaddr
551 in the totem section of the config to an address of the new network. It can be
552 any address from the subnet configured on the new network interface.
554 After you increased the 'config_version' property the new configuration file
570 ring0_addr: 10.10.10.2
577 ring0_addr: 10.10.10.3
584 ring0_addr: 10.10.10.1
590 provider: corosync_votequorum
594 cluster_name: thomas-testcluster
600 bindnetaddr: 10.10.10.1
607 Now after a final check whether all changed information is correct we save it
608 and see again the <<edit-corosync-conf,edit corosync.conf file>> section to
609 learn how to bring it in effect.
611 As our change cannot be enforced live from corosync we have to do an restart.
613 On a single node execute:
616 systemctl restart corosync
619 Now check if everything is fine:
623 systemctl status corosync
626 If corosync runs again correct restart corosync also on all other nodes.
627 They will then join the cluster membership one by one on the new network.
629 Redundant Ring Protocol
630 ~~~~~~~~~~~~~~~~~~~~~~~
631 To avoid a single point of failure you should implement counter measurements.
632 This can be on the hardware and operating system level through network bonding.
634 Corosync itself offers also a possibility to add redundancy through the so
635 called 'Redundant Ring Protocol'. This protocol allows running a second totem
636 ring on another network, this network should be physically separated from the
637 other rings network to actually increase availability.
639 RRP On Cluster Creation
640 ~~~~~~~~~~~~~~~~~~~~~~~
642 The 'pvecm create' command provides the additional parameters 'bindnetX_addr',
643 'ringX_addr' and 'rrp_mode', can be used for RRP configuration.
645 NOTE: See the <<corosync-conf-glossary,glossary>> if you do not know what each parameter means.
647 So if you have two networks, one on the 10.10.10.1/24 and the other on the
648 10.10.20.1/24 subnet you would execute:
652 pvecm create CLUSTERNAME -bindnet0_addr 10.10.10.1 -ring0_addr 10.10.10.1 \
653 -bindnet1_addr 10.10.20.1 -ring1_addr 10.10.20.1
656 RRP On Existing Clusters
657 ~~~~~~~~~~~~~~~~~~~~~~~~
659 You will take similar steps as described in
660 <<separate-cluster-net-after-creation,separating the cluster network>> to
661 enable RRP on an already running cluster. The single difference is, that you
662 will add `ring1` and use it instead of `ring0`.
664 First add a new `interface` subsection in the `totem` section, set its
665 `ringnumber` property to `1`. Set the interfaces `bindnetaddr` property to an
666 address of the subnet you have configured for your new ring.
667 Further set the `rrp_mode` to `passive`, this is the only stable mode.
669 Then add to each node entry in the `nodelist` section its new `ring1_addr`
670 property with the nodes additional ring address.
672 So if you have two networks, one on the 10.10.10.1/24 and the other on the
673 10.10.20.1/24 subnet, the final configuration file should look like:
684 bindnetaddr: 10.10.10.1
688 bindnetaddr: 10.10.20.1
698 ring0_addr: 10.10.10.1
699 ring1_addr: 10.10.20.1
706 ring0_addr: 10.10.10.2
707 ring1_addr: 10.10.20.2
710 [...] # other cluster nodes here
713 [...] # other remaining config sections here
717 Bring it in effect like described in the
718 <<edit-corosync-conf,edit the corosync.conf file>> section.
720 This is a change which cannot take live in effect and needs at least a restart
721 of corosync. Recommended is a restart of the whole cluster.
723 If you cannot reboot the whole cluster ensure no High Availability services are
724 configured and the stop the corosync service on all nodes. After corosync is
725 stopped on all nodes start it one after the other again.
727 Corosync Configuration
728 ----------------------
730 The `/etc/pve/corosync.conf` file plays a central role in {pve} cluster. It
731 controls the cluster member ship and its network.
732 For reading more about it check the corosync.conf man page:
738 For node membership you should always use the `pvecm` tool provided by {pve}.
739 You may have to edit the configuration file manually for other changes.
740 Here are a few best practice tips for doing this.
742 [[edit-corosync-conf]]
746 Editing the corosync.conf file can be not always straight forward. There are
747 two on each cluster, one in `/etc/pve/corosync.conf` and the other in
748 `/etc/corosync/corosync.conf`. Editing the one in our cluster file system will
749 propagate the changes to the local one, but not vice versa.
751 The configuration will get updated automatically as soon as the file changes.
752 This means changes which can be integrated in a running corosync will take
753 instantly effect. So you should always make a copy and edit that instead, to
754 avoid triggering some unwanted changes by an in between safe.
758 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
761 Then open the Config file with your favorite editor, `nano` and `vim.tiny` are
762 preinstalled on {pve} for example.
764 NOTE: Always increment the 'config_version' number on configuration changes,
765 omitting this can lead to problems.
767 After making the necessary changes create another copy of the current working
768 configuration file. This serves as a backup if the new configuration fails to
769 apply or makes problems in other ways.
773 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak
776 Then move the new configuration file over the old one:
779 mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
782 You may check with the commands
785 systemctl status corosync
786 journalctl -b -u corosync
789 If the change could applied automatically. If not you may have to restart the
790 corosync service via:
793 systemctl restart corosync
796 On errors check the troubleshooting section below.
801 Issue: 'quorum.expected_votes must be configured'
802 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
804 When corosync starts to fail and you get the following message in the system log:
808 corosync[1647]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
809 corosync[1647]: [SERV ] Service engine 'corosync_quorum' failed to load for reason
810 'configuration error: nodelist or quorum.expected_votes must be configured!'
814 It means that the hostname you set for corosync 'ringX_addr' in the
815 configuration could not be resolved.
818 Write Configuration When Not Quorate
819 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
821 If you need to change '/etc/pve/corosync.conf' on an node with no quorum, and you
822 know what you do, use:
828 This sets the expected vote count to 1 and makes the cluster quorate. You can
829 now fix your configuration, or revert it back to the last working backup.
831 This is not enough if corosync cannot start anymore. Here its best to edit the
832 local copy of the corosync configuration in '/etc/corosync/corosync.conf' so
833 that corosync can start again. Ensure that on all nodes this configuration has
834 the same content to avoid split brains. If you are not sure what went wrong
835 it's best to ask the Proxmox Community to help you.
838 [[corosync-conf-glossary]]
839 Corosync Configuration Glossary
840 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
843 This names the different ring addresses for the corosync totem rings used for
844 the cluster communication.
847 Defines to which interface the ring should bind to. It may be any address of
848 the subnet configured on the interface we want to use. In general its the
849 recommended to just use an address a node uses on this interface.
852 Specifies the mode of the redundant ring protocol and may be passive, active or
853 none. Note that use of active is highly experimental and not official
854 supported. Passive is the preferred mode, it may double the cluster
855 communication throughput and increases availability.
861 It is obvious that a cluster is not quorate when all nodes are
862 offline. This is a common case after a power failure.
864 NOTE: It is always a good idea to use an uninterruptible power supply
865 (``UPS'', also called ``battery backup'') to avoid this state, especially if
868 On node startup, the `pve-guests` service is started and waits for
869 quorum. Once quorate, it starts all guests which have the `onboot`
872 When you turn on nodes, or when power comes back after power failure,
873 it is likely that some nodes boots faster than others. Please keep in
874 mind that guest startup is delayed until you reach quorum.
880 Migrating virtual guests to other nodes is a useful feature in a
881 cluster. There are settings to control the behavior of such
882 migrations. This can be done via the configuration file
883 `datacenter.cfg` or for a specific migration via API or command line
886 It makes a difference if a Guest is online or offline, or if it has
887 local resources (like a local disk).
889 For Details about Virtual Machine Migration see the
890 xref:qm_migration[QEMU/KVM Migration Chapter]
892 For Details about Container Migration see the
893 xref:pct_migration[Container Migration Chapter]
898 The migration type defines if the migration data should be sent over an
899 encrypted (`secure`) channel or an unencrypted (`insecure`) one.
900 Setting the migration type to insecure means that the RAM content of a
901 virtual guest gets also transferred unencrypted, which can lead to
902 information disclosure of critical data from inside the guest (for
903 example passwords or encryption keys).
905 Therefore, we strongly recommend using the secure channel if you do
906 not have full control over the network and can not guarantee that no
907 one is eavesdropping to it.
909 NOTE: Storage migration does not follow this setting. Currently, it
910 always sends the storage content over a secure channel.
912 Encryption requires a lot of computing power, so this setting is often
913 changed to "unsafe" to achieve better performance. The impact on
914 modern systems is lower because they implement AES encryption in
915 hardware. The performance impact is particularly evident in fast
916 networks where you can transfer 10 Gbps or more.
922 By default, {pve} uses the network in which cluster communication
923 takes place to send the migration traffic. This is not optimal because
924 sensitive cluster traffic can be disrupted and this network may not
925 have the best bandwidth available on the node.
927 Setting the migration network parameter allows the use of a dedicated
928 network for the entire migration traffic. In addition to the memory,
929 this also affects the storage traffic for offline migrations.
931 The migration network is set as a network in the CIDR notation. This
932 has the advantage that you do not have to set individual IP addresses
933 for each node. {pve} can determine the real address on the
934 destination node from the network specified in the CIDR form. To
935 enable this, the network must be specified so that each node has one,
936 but only one IP in the respective network.
942 We assume that we have a three-node setup with three separate
943 networks. One for public communication with the Internet, one for
944 cluster communication and a very fast one, which we want to use as a
945 dedicated network for migration.
947 A network configuration for such a setup might look as follows:
950 iface eno1 inet manual
954 iface vmbr0 inet static
956 netmask 255.255.250.0
964 iface eno2 inet static
966 netmask 255.255.255.0
970 iface eno3 inet static
972 netmask 255.255.255.0
975 Here, we will use the network 10.1.2.0/24 as a migration network. For
976 a single migration, you can do this using the `migration_network`
977 parameter of the command line tool:
980 # qm migrate 106 tre --online --migration_network 10.1.2.0/24
983 To configure this as the default network for all migrations in the
984 cluster, set the `migration` property of the `/etc/pve/datacenter.cfg`
988 # use dedicated migration network
989 migration: secure,network=10.1.2.0/24
992 NOTE: The migration type must always be set when the migration network
993 gets set in `/etc/pve/datacenter.cfg`.
997 include::pve-copyright.adoc[]