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 * Root password of a cluster node is required for adding nodes.
80 NOTE: It is not possible to mix {pve} 3.x and earlier with {pve} 4.X cluster
83 NOTE: While it's possible for {pve} 4.4 and {pve} 5.0 this is not supported as
84 production configuration and should only used temporarily during upgrading the
85 whole cluster from one to another major version.
91 First, install {PVE} on all nodes. Make sure that each node is
92 installed with the final hostname and IP configuration. Changing the
93 hostname and IP is not possible after cluster creation.
95 Currently the cluster creation can either be done on the console (login via
96 `ssh`) or the API, which we have a GUI implementation for (__Datacenter ->
99 While it's often common use to reference all other nodenames in `/etc/hosts`
100 with their IP this is not strictly necessary for a cluster, which normally uses
101 multicast, to work. It maybe useful as you then can connect from one node to
102 the other with SSH through the easier to remember node name.
104 [[pvecm_create_cluster]]
108 Login via `ssh` to the first {pve} node. Use a unique name for your cluster.
109 This name cannot be changed later. The cluster name follows the same rules as
113 hp1# pvecm create CLUSTERNAME
116 CAUTION: The cluster name is used to compute the default multicast address.
117 Please use unique cluster names if you run more than one cluster inside your
118 network. To avoid human confusion, it is also recommended to choose different
119 names even if clusters do not share the cluster network.
121 To check the state of your cluster use:
127 Multiple Clusters In Same Network
128 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
130 It is possible to create multiple clusters in the same physical or logical
131 network. Each cluster must have a unique name, which is used to generate the
132 cluster's multicast group address. As long as no duplicate cluster names are
133 configured in one network segment, the different clusters won't interfere with
136 If multiple clusters operate in a single network it may be beneficial to setup
137 an IGMP querier and enable IGMP Snooping in said network. This may reduce the
138 load of the network significantly because multicast packets are only delivered
139 to endpoints of the respective member nodes.
142 [[pvecm_join_node_to_cluster]]
143 Adding Nodes to the Cluster
144 ---------------------------
146 Login via `ssh` to the node you want to add.
149 hp2# pvecm add IP-ADDRESS-CLUSTER
152 For `IP-ADDRESS-CLUSTER` use the IP from an existing cluster node.
154 CAUTION: A new node cannot hold any VMs, because you would get
155 conflicts about identical VM IDs. Also, all existing configuration in
156 `/etc/pve` is overwritten when you join a new node to the cluster. To
157 workaround, use `vzdump` to backup and restore to a different VMID after
158 adding the node to the cluster.
160 To check the state of cluster:
166 .Cluster status after adding 4 nodes
171 Date: Mon Apr 20 12:30:13 2015
172 Quorum provider: corosync_votequorum
178 Votequorum information
179 ~~~~~~~~~~~~~~~~~~~~~~
186 Membership information
187 ~~~~~~~~~~~~~~~~~~~~~~
189 0x00000001 1 192.168.15.91
190 0x00000002 1 192.168.15.92 (local)
191 0x00000003 1 192.168.15.93
192 0x00000004 1 192.168.15.94
195 If you only want the list of all nodes use:
201 .List nodes in a cluster
205 Membership information
206 ~~~~~~~~~~~~~~~~~~~~~~
214 [[adding-nodes-with-separated-cluster-network]]
215 Adding Nodes With Separated Cluster Network
216 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
218 When adding a node to a cluster with a separated cluster network you need to
219 use the 'ringX_addr' parameters to set the nodes address on those networks:
223 pvecm add IP-ADDRESS-CLUSTER -ring0_addr IP-ADDRESS-RING0
226 If you want to use the Redundant Ring Protocol you will also want to pass the
227 'ring1_addr' parameter.
230 Remove a Cluster Node
231 ---------------------
233 CAUTION: Read carefully the procedure before proceeding, as it could
234 not be what you want or need.
236 Move all virtual machines from the node. Make sure you have no local
237 data or backups you want to keep, or save them accordingly.
238 In the following example we will remove the node hp4 from the cluster.
240 Log in to a *different* cluster node (not hp4), and issue a `pvecm nodes`
241 command to identify the node ID to remove:
246 Membership information
247 ~~~~~~~~~~~~~~~~~~~~~~
256 At this point you must power off hp4 and
257 make sure that it will not power on again (in the network) as it
260 IMPORTANT: As said above, it is critical to power off the node
261 *before* removal, and make sure that it will *never* power on again
262 (in the existing cluster network) as it is.
263 If you power on the node as it is, your cluster will be screwed up and
264 it could be difficult to restore a clean cluster state.
266 After powering off the node hp4, we can safely remove it from the cluster.
269 hp1# pvecm delnode hp4
272 If the operation succeeds no output is returned, just check the node
273 list again with `pvecm nodes` or `pvecm status`. You should see
281 Date: Mon Apr 20 12:44:28 2015
282 Quorum provider: corosync_votequorum
288 Votequorum information
289 ~~~~~~~~~~~~~~~~~~~~~~
296 Membership information
297 ~~~~~~~~~~~~~~~~~~~~~~
299 0x00000001 1 192.168.15.90 (local)
300 0x00000002 1 192.168.15.91
301 0x00000003 1 192.168.15.92
304 If, for whatever reason, you want that this server joins the same
305 cluster again, you have to
307 * reinstall {pve} on it from scratch
309 * then join it, as explained in the previous section.
311 [[pvecm_separate_node_without_reinstall]]
312 Separate A Node Without Reinstalling
313 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
315 CAUTION: This is *not* the recommended method, proceed with caution. Use the
316 above mentioned method if you're unsure.
318 You can also separate a node from a cluster without reinstalling it from
319 scratch. But after removing the node from the cluster it will still have
320 access to the shared storages! This must be resolved before you start removing
321 the node from the cluster. A {pve} cluster cannot share the exact same
322 storage with another cluster, as storage locking doesn't work over cluster
323 boundary. Further, it may also lead to VMID conflicts.
325 Its suggested that you create a new storage where only the node which you want
326 to separate has access. This can be an new export on your NFS or a new Ceph
327 pool, to name a few examples. Its just important that the exact same storage
328 does not gets accessed by multiple clusters. After setting this storage up move
329 all data from the node and its VMs to it. Then you are ready to separate the
330 node from the cluster.
332 WARNING: Ensure all shared resources are cleanly separated! You will run into
333 conflicts and problems else.
335 First stop the corosync and the pve-cluster services on the node:
338 systemctl stop pve-cluster
339 systemctl stop corosync
342 Start the cluster filesystem again in local mode:
348 Delete the corosync configuration files:
351 rm /etc/pve/corosync.conf
355 You can now start the filesystem again as normal service:
359 systemctl start pve-cluster
362 The node is now separated from the cluster. You can deleted it from a remaining
363 node of the cluster with:
366 pvecm delnode oldnode
369 If the command failed, because the remaining node in the cluster lost quorum
370 when the now separate node exited, you may set the expected votes to 1 as a workaround:
376 And then repeat the 'pvecm delnode' command.
378 Now switch back to the separated node, here delete all remaining files left
379 from the old cluster. This ensures that the node can be added to another
380 cluster again without problems.
384 rm /var/lib/corosync/*
387 As the configuration files from the other nodes are still in the cluster
388 filesystem you may want to clean those up too. Remove simply the whole
389 directory recursive from '/etc/pve/nodes/NODENAME', but check three times that
390 you used the correct one before deleting it.
392 CAUTION: The nodes SSH keys are still in the 'authorized_key' file, this means
393 the nodes can still connect to each other with public key authentication. This
394 should be fixed by removing the respective keys from the
395 '/etc/pve/priv/authorized_keys' file.
400 {pve} use a quorum-based technique to provide a consistent state among
403 [quote, from Wikipedia, Quorum (distributed computing)]
405 A quorum is the minimum number of votes that a distributed transaction
406 has to obtain in order to be allowed to perform an operation in a
410 In case of network partitioning, state changes requires that a
411 majority of nodes are online. The cluster switches to read-only mode
414 NOTE: {pve} assigns a single vote to each node by default.
419 The cluster network is the core of a cluster. All messages sent over it have to
420 be delivered reliable to all nodes in their respective order. In {pve} this
421 part is done by corosync, an implementation of a high performance low overhead
422 high availability development toolkit. It serves our decentralized
423 configuration file system (`pmxcfs`).
425 [[cluster-network-requirements]]
428 This needs a reliable network with latencies under 2 milliseconds (LAN
429 performance) to work properly. While corosync can also use unicast for
430 communication between nodes its **highly recommended** to have a multicast
431 capable network. The network should not be used heavily by other members,
432 ideally corosync runs on its own network.
433 *never* share it with network where storage communicates too.
435 Before setting up a cluster it is good practice to check if the network is fit
438 * Ensure that all nodes are in the same subnet. This must only be true for the
439 network interfaces used for cluster communication (corosync).
441 * Ensure all nodes can reach each other over those interfaces, using `ping` is
442 enough for a basic test.
444 * Ensure that multicast works in general and a high package rates. This can be
445 done with the `omping` tool. The final "%loss" number should be < 1%.
449 omping -c 10000 -i 0.001 -F -q NODE1-IP NODE2-IP ...
452 * Ensure that multicast communication works over an extended period of time.
453 This uncovers problems where IGMP snooping is activated on the network but
454 no multicast querier is active. This test has a duration of around 10
459 omping -c 600 -i 1 -q NODE1-IP NODE2-IP ...
462 Your network is not ready for clustering if any of these test fails. Recheck
463 your network configuration. Especially switches are notorious for having
464 multicast disabled by default or IGMP snooping enabled with no IGMP querier
467 In smaller cluster its also an option to use unicast if you really cannot get
470 Separate Cluster Network
471 ~~~~~~~~~~~~~~~~~~~~~~~~
473 When creating a cluster without any parameters the cluster network is generally
474 shared with the Web UI and the VMs and its traffic. Depending on your setup
475 even storage traffic may get sent over the same network. Its recommended to
476 change that, as corosync is a time critical real time application.
478 Setting Up A New Network
479 ^^^^^^^^^^^^^^^^^^^^^^^^
481 First you have to setup a new network interface. It should be on a physical
482 separate network. Ensure that your network fulfills the
483 <<cluster-network-requirements,cluster network requirements>>.
485 Separate On Cluster Creation
486 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
488 This is possible through the 'ring0_addr' and 'bindnet0_addr' parameter of
489 the 'pvecm create' command used for creating a new cluster.
491 If you have setup an additional NIC with a static address on 10.10.10.1/25
492 and want to send and receive all cluster communication over this interface
497 pvecm create test --ring0_addr 10.10.10.1 --bindnet0_addr 10.10.10.0
500 To check if everything is working properly execute:
503 systemctl status corosync
506 Afterwards, proceed as descripted in the section to
507 <<adding-nodes-with-separated-cluster-network,add nodes with a separated cluster network>>.
509 [[separate-cluster-net-after-creation]]
510 Separate After Cluster Creation
511 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
513 You can do this also if you have already created a cluster and want to switch
514 its communication to another network, without rebuilding the whole cluster.
515 This change may lead to short durations of quorum loss in the cluster, as nodes
516 have to restart corosync and come up one after the other on the new network.
518 Check how to <<edit-corosync-conf,edit the corosync.conf file>> first.
519 The open it and you should see a file similar to:
553 provider: corosync_votequorum
557 cluster_name: thomas-testcluster
563 bindnetaddr: 192.168.30.50
570 The first you want to do is add the 'name' properties in the node entries if
571 you do not see them already. Those *must* match the node name.
573 Then replace the address from the 'ring0_addr' properties with the new
574 addresses. You may use plain IP addresses or also hostnames here. If you use
575 hostnames ensure that they are resolvable from all nodes.
577 In my example I want to switch my cluster communication to the 10.10.10.1/25
578 network. So I replace all 'ring0_addr' respectively. I also set the bindnetaddr
579 in the totem section of the config to an address of the new network. It can be
580 any address from the subnet configured on the new network interface.
582 After you increased the 'config_version' property the new configuration file
598 ring0_addr: 10.10.10.2
605 ring0_addr: 10.10.10.3
612 ring0_addr: 10.10.10.1
618 provider: corosync_votequorum
622 cluster_name: thomas-testcluster
628 bindnetaddr: 10.10.10.1
635 Now after a final check whether all changed information is correct we save it
636 and see again the <<edit-corosync-conf,edit corosync.conf file>> section to
637 learn how to bring it in effect.
639 As our change cannot be enforced live from corosync we have to do an restart.
641 On a single node execute:
644 systemctl restart corosync
647 Now check if everything is fine:
651 systemctl status corosync
654 If corosync runs again correct restart corosync also on all other nodes.
655 They will then join the cluster membership one by one on the new network.
658 Redundant Ring Protocol
659 ~~~~~~~~~~~~~~~~~~~~~~~
660 To avoid a single point of failure you should implement counter measurements.
661 This can be on the hardware and operating system level through network bonding.
663 Corosync itself offers also a possibility to add redundancy through the so
664 called 'Redundant Ring Protocol'. This protocol allows running a second totem
665 ring on another network, this network should be physically separated from the
666 other rings network to actually increase availability.
668 RRP On Cluster Creation
669 ~~~~~~~~~~~~~~~~~~~~~~~
671 The 'pvecm create' command provides the additional parameters 'bindnetX_addr',
672 'ringX_addr' and 'rrp_mode', can be used for RRP configuration.
674 NOTE: See the <<corosync-conf-glossary,glossary>> if you do not know what each parameter means.
676 So if you have two networks, one on the 10.10.10.1/24 and the other on the
677 10.10.20.1/24 subnet you would execute:
681 pvecm create CLUSTERNAME -bindnet0_addr 10.10.10.1 -ring0_addr 10.10.10.1 \
682 -bindnet1_addr 10.10.20.1 -ring1_addr 10.10.20.1
685 RRP On Existing Clusters
686 ~~~~~~~~~~~~~~~~~~~~~~~~
688 You will take similar steps as described in
689 <<separate-cluster-net-after-creation,separating the cluster network>> to
690 enable RRP on an already running cluster. The single difference is, that you
691 will add `ring1` and use it instead of `ring0`.
693 First add a new `interface` subsection in the `totem` section, set its
694 `ringnumber` property to `1`. Set the interfaces `bindnetaddr` property to an
695 address of the subnet you have configured for your new ring.
696 Further set the `rrp_mode` to `passive`, this is the only stable mode.
698 Then add to each node entry in the `nodelist` section its new `ring1_addr`
699 property with the nodes additional ring address.
701 So if you have two networks, one on the 10.10.10.1/24 and the other on the
702 10.10.20.1/24 subnet, the final configuration file should look like:
713 bindnetaddr: 10.10.10.1
717 bindnetaddr: 10.10.20.1
727 ring0_addr: 10.10.10.1
728 ring1_addr: 10.10.20.1
735 ring0_addr: 10.10.10.2
736 ring1_addr: 10.10.20.2
739 [...] # other cluster nodes here
742 [...] # other remaining config sections here
746 Bring it in effect like described in the
747 <<edit-corosync-conf,edit the corosync.conf file>> section.
749 This is a change which cannot take live in effect and needs at least a restart
750 of corosync. Recommended is a restart of the whole cluster.
752 If you cannot reboot the whole cluster ensure no High Availability services are
753 configured and the stop the corosync service on all nodes. After corosync is
754 stopped on all nodes start it one after the other again.
756 Corosync Configuration
757 ----------------------
759 The `/etc/pve/corosync.conf` file plays a central role in {pve} cluster. It
760 controls the cluster member ship and its network.
761 For reading more about it check the corosync.conf man page:
767 For node membership you should always use the `pvecm` tool provided by {pve}.
768 You may have to edit the configuration file manually for other changes.
769 Here are a few best practice tips for doing this.
771 [[edit-corosync-conf]]
775 Editing the corosync.conf file can be not always straight forward. There are
776 two on each cluster, one in `/etc/pve/corosync.conf` and the other in
777 `/etc/corosync/corosync.conf`. Editing the one in our cluster file system will
778 propagate the changes to the local one, but not vice versa.
780 The configuration will get updated automatically as soon as the file changes.
781 This means changes which can be integrated in a running corosync will take
782 instantly effect. So you should always make a copy and edit that instead, to
783 avoid triggering some unwanted changes by an in between safe.
787 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
790 Then open the Config file with your favorite editor, `nano` and `vim.tiny` are
791 preinstalled on {pve} for example.
793 NOTE: Always increment the 'config_version' number on configuration changes,
794 omitting this can lead to problems.
796 After making the necessary changes create another copy of the current working
797 configuration file. This serves as a backup if the new configuration fails to
798 apply or makes problems in other ways.
802 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak
805 Then move the new configuration file over the old one:
808 mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
811 You may check with the commands
814 systemctl status corosync
815 journalctl -b -u corosync
818 If the change could applied automatically. If not you may have to restart the
819 corosync service via:
822 systemctl restart corosync
825 On errors check the troubleshooting section below.
830 Issue: 'quorum.expected_votes must be configured'
831 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
833 When corosync starts to fail and you get the following message in the system log:
837 corosync[1647]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
838 corosync[1647]: [SERV ] Service engine 'corosync_quorum' failed to load for reason
839 'configuration error: nodelist or quorum.expected_votes must be configured!'
843 It means that the hostname you set for corosync 'ringX_addr' in the
844 configuration could not be resolved.
847 Write Configuration When Not Quorate
848 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
850 If you need to change '/etc/pve/corosync.conf' on an node with no quorum, and you
851 know what you do, use:
857 This sets the expected vote count to 1 and makes the cluster quorate. You can
858 now fix your configuration, or revert it back to the last working backup.
860 This is not enough if corosync cannot start anymore. Here its best to edit the
861 local copy of the corosync configuration in '/etc/corosync/corosync.conf' so
862 that corosync can start again. Ensure that on all nodes this configuration has
863 the same content to avoid split brains. If you are not sure what went wrong
864 it's best to ask the Proxmox Community to help you.
867 [[corosync-conf-glossary]]
868 Corosync Configuration Glossary
869 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
872 This names the different ring addresses for the corosync totem rings used for
873 the cluster communication.
876 Defines to which interface the ring should bind to. It may be any address of
877 the subnet configured on the interface we want to use. In general its the
878 recommended to just use an address a node uses on this interface.
881 Specifies the mode of the redundant ring protocol and may be passive, active or
882 none. Note that use of active is highly experimental and not official
883 supported. Passive is the preferred mode, it may double the cluster
884 communication throughput and increases availability.
890 It is obvious that a cluster is not quorate when all nodes are
891 offline. This is a common case after a power failure.
893 NOTE: It is always a good idea to use an uninterruptible power supply
894 (``UPS'', also called ``battery backup'') to avoid this state, especially if
897 On node startup, the `pve-guests` service is started and waits for
898 quorum. Once quorate, it starts all guests which have the `onboot`
901 When you turn on nodes, or when power comes back after power failure,
902 it is likely that some nodes boots faster than others. Please keep in
903 mind that guest startup is delayed until you reach quorum.
909 Migrating virtual guests to other nodes is a useful feature in a
910 cluster. There are settings to control the behavior of such
911 migrations. This can be done via the configuration file
912 `datacenter.cfg` or for a specific migration via API or command line
915 It makes a difference if a Guest is online or offline, or if it has
916 local resources (like a local disk).
918 For Details about Virtual Machine Migration see the
919 xref:qm_migration[QEMU/KVM Migration Chapter]
921 For Details about Container Migration see the
922 xref:pct_migration[Container Migration Chapter]
927 The migration type defines if the migration data should be sent over an
928 encrypted (`secure`) channel or an unencrypted (`insecure`) one.
929 Setting the migration type to insecure means that the RAM content of a
930 virtual guest gets also transferred unencrypted, which can lead to
931 information disclosure of critical data from inside the guest (for
932 example passwords or encryption keys).
934 Therefore, we strongly recommend using the secure channel if you do
935 not have full control over the network and can not guarantee that no
936 one is eavesdropping to it.
938 NOTE: Storage migration does not follow this setting. Currently, it
939 always sends the storage content over a secure channel.
941 Encryption requires a lot of computing power, so this setting is often
942 changed to "unsafe" to achieve better performance. The impact on
943 modern systems is lower because they implement AES encryption in
944 hardware. The performance impact is particularly evident in fast
945 networks where you can transfer 10 Gbps or more.
951 By default, {pve} uses the network in which cluster communication
952 takes place to send the migration traffic. This is not optimal because
953 sensitive cluster traffic can be disrupted and this network may not
954 have the best bandwidth available on the node.
956 Setting the migration network parameter allows the use of a dedicated
957 network for the entire migration traffic. In addition to the memory,
958 this also affects the storage traffic for offline migrations.
960 The migration network is set as a network in the CIDR notation. This
961 has the advantage that you do not have to set individual IP addresses
962 for each node. {pve} can determine the real address on the
963 destination node from the network specified in the CIDR form. To
964 enable this, the network must be specified so that each node has one,
965 but only one IP in the respective network.
971 We assume that we have a three-node setup with three separate
972 networks. One for public communication with the Internet, one for
973 cluster communication and a very fast one, which we want to use as a
974 dedicated network for migration.
976 A network configuration for such a setup might look as follows:
979 iface eno1 inet manual
983 iface vmbr0 inet static
985 netmask 255.255.250.0
993 iface eno2 inet static
995 netmask 255.255.255.0
999 iface eno3 inet static
1001 netmask 255.255.255.0
1004 Here, we will use the network 10.1.2.0/24 as a migration network. For
1005 a single migration, you can do this using the `migration_network`
1006 parameter of the command line tool:
1009 # qm migrate 106 tre --online --migration_network 10.1.2.0/24
1012 To configure this as the default network for all migrations in the
1013 cluster, set the `migration` property of the `/etc/pve/datacenter.cfg`
1017 # use dedicated migration network
1018 migration: secure,network=10.1.2.0/24
1021 NOTE: The migration type must always be set when the migration network
1022 gets set in `/etc/pve/datacenter.cfg`.
1026 include::pve-copyright.adoc[]