9 pvecm - Proxmox VE Cluster Manager
14 include::pvecm.1-synopsis.adoc[]
26 The {PVE} cluster manager `pvecm` is a tool to create a group of
27 physical servers. Such a group is called a *cluster*. We use the
28 http://www.corosync.org[Corosync Cluster Engine] for reliable group
29 communication, and such clusters can consist of up to 32 physical nodes
30 (probably more, dependent on network latency).
32 `pvecm` can be used to create a new cluster, join nodes to a cluster,
33 leave the cluster, get status information and do various other cluster
34 related tasks. The **P**rox**m**o**x** **C**luster **F**ile **S**ystem (``pmxcfs'')
35 is used to transparently distribute the cluster configuration to all cluster
38 Grouping nodes into a cluster has the following advantages:
40 * Centralized, web based management
42 * Multi-master clusters: each node can do all management task
44 * `pmxcfs`: database-driven file system for storing configuration files,
45 replicated in real-time on all nodes using `corosync`.
47 * Easy migration of virtual machines and containers between physical
52 * Cluster-wide services like firewall and HA
58 * All nodes must be in the same network as `corosync` uses IP Multicast
59 to communicate between nodes (also see
60 http://www.corosync.org[Corosync Cluster Engine]). Corosync uses UDP
61 ports 5404 and 5405 for cluster communication.
63 NOTE: Some switches do not support IP multicast by default and must be
64 manually enabled first.
66 * Date and time have to be synchronized.
68 * SSH tunnel on TCP port 22 between nodes is used.
70 * If you are interested in High Availability, you need to have at
71 least three nodes for reliable quorum. All nodes should have the
74 * We recommend a dedicated NIC for the cluster traffic, especially if
75 you use shared storage.
77 NOTE: It is not possible to mix Proxmox VE 3.x and earlier with
78 Proxmox VE 4.0 cluster nodes.
84 First, install {PVE} on all nodes. Make sure that each node is
85 installed with the final hostname and IP configuration. Changing the
86 hostname and IP is not possible after cluster creation.
88 Currently the cluster creation has to be done on the console, so you
89 need to login via `ssh`.
94 Login via `ssh` to the first {pve} node. Use a unique name for your cluster.
95 This name cannot be changed later.
97 hp1# pvecm create YOUR-CLUSTER-NAME
99 CAUTION: The cluster name is used to compute the default multicast
100 address. Please use unique cluster names if you run more than one
101 cluster inside your network.
103 To check the state of your cluster use:
108 Adding Nodes to the Cluster
109 ---------------------------
111 Login via `ssh` to the node you want to add.
113 hp2# pvecm add IP-ADDRESS-CLUSTER
115 For `IP-ADDRESS-CLUSTER` use the IP from an existing cluster node.
117 CAUTION: A new node cannot hold any VMs, because you would get
118 conflicts about identical VM IDs. Also, all existing configuration in
119 `/etc/pve` is overwritten when you join a new node to the cluster. To
120 workaround, use `vzdump` to backup and restore to a different VMID after
121 adding the node to the cluster.
123 To check the state of cluster:
127 .Cluster status after adding 4 nodes
132 Date: Mon Apr 20 12:30:13 2015
133 Quorum provider: corosync_votequorum
139 Votequorum information
140 ~~~~~~~~~~~~~~~~~~~~~~
147 Membership information
148 ~~~~~~~~~~~~~~~~~~~~~~
150 0x00000001 1 192.168.15.91
151 0x00000002 1 192.168.15.92 (local)
152 0x00000003 1 192.168.15.93
153 0x00000004 1 192.168.15.94
156 If you only want the list of all nodes use:
160 .List nodes in a cluster
164 Membership information
165 ~~~~~~~~~~~~~~~~~~~~~~
173 Adding Nodes With Separated Cluster Network
174 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
176 When adding a node to a cluster with a separated cluster network you need to
177 use the 'ringX_addr' parameters to set the nodes address on those networks:
181 pvecm add IP-ADDRESS-CLUSTER -ring0_addr IP-ADDRESS-RING0
184 If you want to use the Redundant Ring Protocol you will also want to pass the
185 'ring1_addr' parameter.
188 Remove a Cluster Node
189 ---------------------
191 CAUTION: Read carefully the procedure before proceeding, as it could
192 not be what you want or need.
194 Move all virtual machines from the node. Make sure you have no local
195 data or backups you want to keep, or save them accordingly.
196 In the following example we will remove the node hp4 from the cluster.
198 Log in to a *different* cluster node (not hp4), and issue a `pvecm nodes`
199 command to identify the node ID to remove:
204 Membership information
205 ~~~~~~~~~~~~~~~~~~~~~~
214 At this point you must power off hp4 and
215 make sure that it will not power on again (in the network) as it
218 IMPORTANT: As said above, it is critical to power off the node
219 *before* removal, and make sure that it will *never* power on again
220 (in the existing cluster network) as it is.
221 If you power on the node as it is, your cluster will be screwed up and
222 it could be difficult to restore a clean cluster state.
224 After powering off the node hp4, we can safely remove it from the cluster.
226 hp1# pvecm delnode hp4
228 If the operation succeeds no output is returned, just check the node
229 list again with `pvecm nodes` or `pvecm status`. You should see
237 Date: Mon Apr 20 12:44:28 2015
238 Quorum provider: corosync_votequorum
244 Votequorum information
245 ~~~~~~~~~~~~~~~~~~~~~~
252 Membership information
253 ~~~~~~~~~~~~~~~~~~~~~~
255 0x00000001 1 192.168.15.90 (local)
256 0x00000002 1 192.168.15.91
257 0x00000003 1 192.168.15.92
260 If, for whatever reason, you want that this server joins the same
261 cluster again, you have to
263 * reinstall {pve} on it from scratch
265 * then join it, as explained in the previous section.
267 [[pvecm_separate_node_without_reinstall]]
268 Separate A Node Without Reinstalling
269 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
271 CAUTION: This is *not* the recommended method, proceed with caution. Use the
272 above mentioned method if you're unsure.
274 You can also separate a node from a cluster without reinstalling it from
275 scratch. But after removing the node from the cluster it will still have
276 access to the shared storages! This must be resolved before you start removing
277 the node from the cluster. A {pve} cluster cannot share the exact same
278 storage with another cluster, as it leads to VMID conflicts.
280 Its suggested that you create a new storage where only the node which you want
281 to separate has access. This can be an new export on your NFS or a new Ceph
282 pool, to name a few examples. Its just important that the exact same storage
283 does not gets accessed by multiple clusters. After setting this storage up move
284 all data from the node and its VMs to it. Then you are ready to separate the
285 node from the cluster.
287 WARNING: Ensure all shared resources are cleanly separated! You will run into
288 conflicts and problems else.
290 First stop the corosync and the pve-cluster services on the node:
293 systemctl stop pve-cluster
294 systemctl stop corosync
297 Start the cluster filesystem again in local mode:
303 Delete the corosync configuration files:
306 rm /etc/pve/corosync.conf
310 You can now start the filesystem again as normal service:
314 systemctl start pve-cluster
317 The node is now separated from the cluster. You can deleted it from a remaining
318 node of the cluster with:
321 pvecm delnode oldnode
324 If the command failed, because the remaining node in the cluster lost quorum
325 when the now separate node exited, you may set the expected votes to 1 as a workaround:
331 And the repeat the 'pvecm delnode' command.
333 Now switch back to the separated node, here delete all remaining files left
334 from the old cluster. This ensures that the node can be added to another
335 cluster again without problems.
339 rm /var/lib/corosync/*
342 As the configuration files from the other nodes are still in the cluster
343 filesystem you may want to clean those up too. Remove simply the whole
344 directory recursive from '/etc/pve/nodes/NODENAME', but check three times that
345 you used the correct one before deleting it.
347 CAUTION: The nodes SSH keys are still in the 'authorized_key' file, this means
348 the nodes can still connect to each other with public key authentication. This
349 should be fixed by removing the respective keys from the
350 '/etc/pve/priv/authorized_keys' file.
355 {pve} use a quorum-based technique to provide a consistent state among
358 [quote, from Wikipedia, Quorum (distributed computing)]
360 A quorum is the minimum number of votes that a distributed transaction
361 has to obtain in order to be allowed to perform an operation in a
365 In case of network partitioning, state changes requires that a
366 majority of nodes are online. The cluster switches to read-only mode
369 NOTE: {pve} assigns a single vote to each node by default.
374 The cluster network is the core of a cluster. All messages sent over it have to
375 be delivered reliable to all nodes in their respective order. In {pve} this
376 part is done by corosync, an implementation of a high performance low overhead
377 high availability development toolkit. It serves our decentralized
378 configuration file system (`pmxcfs`).
380 [[cluster-network-requirements]]
383 This needs a reliable network with latencies under 2 milliseconds (LAN
384 performance) to work properly. While corosync can also use unicast for
385 communication between nodes its **highly recommended** to have a multicast
386 capable network. The network should not be used heavily by other members,
387 ideally corosync runs on its own network.
388 *never* share it with network where storage communicates too.
390 Before setting up a cluster it is good practice to check if the network is fit
393 * Ensure that all nodes are in the same subnet. This must only be true for the
394 network interfaces used for cluster communication (corosync).
396 * Ensure all nodes can reach each other over those interfaces, using `ping` is
397 enough for a basic test.
399 * Ensure that multicast works in general and a high package rates. This can be
400 done with the `omping` tool. The final "%loss" number should be < 1%.
403 omping -c 10000 -i 0.001 -F -q NODE1-IP NODE2-IP ...
406 * Ensure that multicast communication works over an extended period of time.
407 This covers up problems where IGMP snooping is activated on the network but
408 no multicast querier is active. This test has a duration of around 10
412 omping -c 600 -i 1 -q NODE1-IP NODE2-IP ...
415 Your network is not ready for clustering if any of these test fails. Recheck
416 your network configuration. Especially switches are notorious for having
417 multicast disabled by default or IGMP snooping enabled with no IGMP querier
420 In smaller cluster its also an option to use unicast if you really cannot get
423 Separate Cluster Network
424 ~~~~~~~~~~~~~~~~~~~~~~~~
426 When creating a cluster without any parameters the cluster network is generally
427 shared with the Web UI and the VMs and its traffic. Depending on your setup
428 even storage traffic may get sent over the same network. Its recommended to
429 change that, as corosync is a time critical real time application.
431 Setting Up A New Network
432 ^^^^^^^^^^^^^^^^^^^^^^^^
434 First you have to setup a new network interface. It should be on a physical
435 separate network. Ensure that your network fulfills the
436 <<cluster-network-requirements,cluster network requirements>>.
438 Separate On Cluster Creation
439 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
441 This is possible through the 'ring0_addr' and 'bindnet0_addr' parameter of
442 the 'pvecm create' command used for creating a new cluster.
444 If you have setup a additional NIC with a static address on 10.10.10.1/25
445 and want to send and receive all cluster communication over this interface
450 pvecm create test --ring0_addr 10.10.10.1 --bindnet0_addr 10.10.10.0
453 To check if everything is working properly execute:
456 systemctl status corosync
459 [[separate-cluster-net-after-creation]]
460 Separate After Cluster Creation
461 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
463 You can do this also if you have already created a cluster and want to switch
464 its communication to another network, without rebuilding the whole cluster.
465 This change may lead to short durations of quorum loss in the cluster, as nodes
466 have to restart corosync and come up one after the other on the new network.
468 Check how to <<edit-corosync-conf,edit the corosync.conf file>> first.
469 The open it and you should see a file similar to:
503 provider: corosync_votequorum
507 cluster_name: thomas-testcluster
513 bindnetaddr: 192.168.30.50
520 The first you want to do is add the 'name' properties in the node entries if
521 you do not see them already. Those *must* match the node name.
523 Then replace the address from the 'ring0_addr' properties with the new
524 addresses. You may use plain IP addresses or also hostnames here. If you use
525 hostnames ensure that they are resolvable from all nodes.
527 In my example I want to switch my cluster communication to the 10.10.10.1/25
528 network. So I replace all 'ring0_addr' respectively. I also set the bindetaddr
529 in the totem section of the config to an address of the new network. It can be
530 any address from the subnet configured on the new network interface.
532 After you increased the 'config_version' property the new configuration file
548 ring0_addr: 10.10.10.2
555 ring0_addr: 10.10.10.3
562 ring0_addr: 10.10.10.1
568 provider: corosync_votequorum
572 cluster_name: thomas-testcluster
578 bindnetaddr: 10.10.10.1
585 Now after a final check whether all changed information is correct we save it
586 and see again the <<edit-corosync-conf,edit corosync.conf file>> section to
587 learn how to bring it in effect.
589 As our change cannot be enforced live from corosync we have to do an restart.
591 On a single node execute:
594 systemctl restart corosync
597 Now check if everything is fine:
601 systemctl status corosync
604 If corosync runs again correct restart corosync also on all other nodes.
605 They will then join the cluster membership one by one on the new network.
607 Redundant Ring Protocol
608 ~~~~~~~~~~~~~~~~~~~~~~~
609 To avoid a single point of failure you should implement counter measurements.
610 This can be on the hardware and operating system level through network bonding.
612 Corosync itself offers also a possibility to add redundancy through the so
613 called 'Redundant Ring Protocol'. This protocol allows running a second totem
614 ring on another network, this network should be physically separated from the
615 other rings network to actually increase availability.
617 RRP On Cluster Creation
618 ~~~~~~~~~~~~~~~~~~~~~~~
620 The 'pvecm create' command provides the additional parameters 'bindnetX_addr',
621 'ringX_addr' and 'rrp_mode', can be used for RRP configuration.
623 NOTE: See the <<corosync-conf-glossary,glossary>> if you do not know what each parameter means.
625 So if you have two networks, one on the 10.10.10.1/24 and the other on the
626 10.10.20.1/24 subnet you would execute:
630 pvecm create CLUSTERNAME -bindnet0_addr 10.10.10.1 -ring0_addr 10.10.10.1 \
631 -bindnet1_addr 10.10.20.1 -ring1_addr 10.10.20.1
634 RRP On A Created Cluster
635 ~~~~~~~~~~~~~~~~~~~~~~~~
637 When enabling an already running cluster to use RRP you will take similar steps
639 <<separate-cluster-net-after-creation,separating the cluster network>>. You
640 just do it on another ring.
642 First add a new `interface` subsection in the `totem` section, set its
643 `ringnumber` property to `1`. Set the interfaces `bindnetaddr` property to an
644 address of the subnet you have configured for your new ring.
645 Further set the `rrp_mode` to `passive`, this is the only stable mode.
647 Then add to each node entry in the `nodelist` section its new `ring1_addr`
648 property with the nodes additional ring address.
650 So if you have two networks, one on the 10.10.10.1/24 and the other on the
651 10.10.20.1/24 subnet, the final configuration file should look like:
662 bindnetaddr: 10.10.10.1
666 bindnetaddr: 10.10.20.1
676 ring0_addr: 10.10.10.1
677 ring1_addr: 10.10.20.1
684 ring0_addr: 10.10.10.2
685 ring1_addr: 10.10.20.2
688 [...] # other cluster nodes here
691 [...] # other remaining config sections here
695 Bring it in effect like described in the
696 <<edit-corosync-conf,edit the corosync.conf file>> section.
698 This is a change which cannot take live in effect and needs at least a restart
699 of corosync. Recommended is a restart of the whole cluster.
701 If you cannot reboot the whole cluster ensure no High Availability services are
702 configured and the stop the corosync service on all nodes. After corosync is
703 stopped on all nodes start it one after the other again.
705 Corosync Configuration
706 ----------------------
708 The `/ect/pve/corosync.conf` file plays a central role in {pve} cluster. It
709 controls the cluster member ship and its network.
710 For reading more about it check the corosync.conf man page:
716 For node membership you should always use the `pvecm` tool provided by {pve}.
717 You may have to edit the configuration file manually for other changes.
718 Here are a few best practice tips for doing this.
720 [[edit-corosync-conf]]
724 Editing the corosync.conf file can be not always straight forward. There are
725 two on each cluster, one in `/etc/pve/corosync.conf` and the other in
726 `/etc/corosync/corosync.conf`. Editing the one in our cluster file system will
727 propagate the changes to the local one, but not vice versa.
729 The configuration will get updated automatically as soon as the file changes.
730 This means changes which can be integrated in a running corosync will take
731 instantly effect. So you should always make a copy and edit that instead, to
732 avoid triggering some unwanted changes by an in between safe.
736 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
739 Then open the Config file with your favorite editor, `nano` and `vim.tiny` are
740 preinstalled on {pve} for example.
742 NOTE: Always increment the 'config_version' number on configuration changes,
743 omitting this can lead to problems.
745 After making the necessary changes create another copy of the current working
746 configuration file. This serves as a backup if the new configuration fails to
747 apply or makes problems in other ways.
751 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak
754 Then move the new configuration file over the old one:
757 mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
760 You may check with the commands
763 systemctl status corosync
764 journalctl -b -u corosync
767 If the change could applied automatically. If not you may have to restart the
768 corosync service via:
771 systemctl restart corosync
774 On errors check the troubleshooting section below.
779 Issue: 'quorum.expected_votes must be configured'
780 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
782 When corosync starts to fail and you get the following message in the system log:
786 corosync[1647]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
787 corosync[1647]: [SERV ] Service engine 'corosync_quorum' failed to load for reason
788 'configuration error: nodelist or quorum.expected_votes must be configured!'
792 It means that the hostname you set for corosync 'ringX_addr' in the
793 configuration could not be resolved.
796 Write Configuration When Not Quorate
797 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
799 If you need to change '/etc/pve/corosync.conf' on an node with no quorum, and you
800 know what you do, use:
806 This sets the expected vote count to 1 and makes the cluster quorate. You can
807 now fix your configuration, or revert it back to the last working backup.
809 This is not enough if corosync cannot start anymore. Here its best to edit the
810 local copy of the corosync configuration in '/etc/corosync/corosync.conf' so
811 that corosync can start again. Ensure that on all nodes this configuration has
812 the same content to avoid split brains. If you are not sure what went wrong
813 it's best to ask the Proxmox Community to help you.
816 [[corosync-conf-glossary]]
817 Corosync Configuration Glossary
818 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
821 This names the different ring addresses for the corosync totem rings used for
822 the cluster communication.
825 Defines to which interface the ring should bind to. It may be any address of
826 the subnet configured on the interface we want to use. In general its the
827 recommended to just use an address a node uses on this interface.
830 Specifies the mode of the redundant ring protocol and may be passive, active or
831 none. Note that use of active is highly experimental and not official
832 supported. Passive is the preferred mode, it may double the cluster
833 communication throughput and increases availability.
839 It is obvious that a cluster is not quorate when all nodes are
840 offline. This is a common case after a power failure.
842 NOTE: It is always a good idea to use an uninterruptible power supply
843 (``UPS'', also called ``battery backup'') to avoid this state, especially if
846 On node startup, service `pve-manager` is started and waits for
847 quorum. Once quorate, it starts all guests which have the `onboot`
850 When you turn on nodes, or when power comes back after power failure,
851 it is likely that some nodes boots faster than others. Please keep in
852 mind that guest startup is delayed until you reach quorum.
858 Migrating virtual guests to other nodes is a useful feature in a
859 cluster. There are settings to control the behavior of such
860 migrations. This can be done via the configuration file
861 `datacenter.cfg` or for a specific migration via API or command line
868 The migration type defines if the migration data should be sent over a
869 encrypted (`secure`) channel or an unencrypted (`insecure`) one.
870 Setting the migration type to insecure means that the RAM content of a
871 virtual guest gets also transfered unencrypted, which can lead to
872 information disclosure of critical data from inside the guest (for
873 example passwords or encryption keys).
875 Therefore, we strongly recommend using the secure channel if you do
876 not have full control over the network and can not guarantee that no
877 one is eavesdropping to it.
879 NOTE: Storage migration does not follow this setting. Currently, it
880 always sends the storage content over a secure channel.
882 Encryption requires a lot of computing power, so this setting is often
883 changed to "unsafe" to achieve better performance. The impact on
884 modern systems is lower because they implement AES encryption in
885 hardware. The performance impact is particularly evident in fast
886 networks where you can transfer 10 Gbps or more.
892 By default, {pve} uses the network in which cluster communication
893 takes place to send the migration traffic. This is not optimal because
894 sensitive cluster traffic can be disrupted and this network may not
895 have the best bandwidth available on the node.
897 Setting the migration network parameter allows the use of a dedicated
898 network for the entire migration traffic. In addition to the memory,
899 this also affects the storage traffic for offline migrations.
901 The migration network is set as a network in the CIDR notation. This
902 has the advantage that you do not have to set individual IP addresses
903 for each node. {pve} can determine the real address on the
904 destination node from the network specified in the CIDR form. To
905 enable this, the network must be specified so that each node has one,
906 but only one IP in the respective network.
912 We assume that we have a three-node setup with three separate
913 networks. One for public communication with the Internet, one for
914 cluster communication and a very fast one, which we want to use as a
915 dedicated network for migration.
917 A network configuration for such a setup might look as follows:
920 iface eth0 inet manual
924 iface vmbr0 inet static
926 netmask 255.255.250.0
934 iface eth1 inet static
936 netmask 255.255.255.0
940 iface eth2 inet static
942 netmask 255.255.255.0
945 Here, we will use the network 10.1.2.0/24 as a migration network. For
946 a single migration, you can do this using the `migration_network`
947 parameter of the command line tool:
950 # qm migrate 106 tre --online --migration_network 10.1.2.0/24
953 To configure this as the default network for all migrations in the
954 cluster, set the `migration` property of the `/etc/pve/datacenter.cfg`
958 # use dedicated migration network
959 migration: secure,network=10.1.2.0/24
962 NOTE: The migration type must always be set when the migration network
963 gets set in `/etc/pve/datacenter.cfg`.
967 include::pve-copyright.adoc[]