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 tasks
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 able to connect to each other via UDP ports 5404 and 5405
62 * Date and time have to be synchronized.
64 * SSH tunnel on TCP port 22 between nodes is used.
66 * If you are interested in High Availability, you need to have at
67 least three nodes for reliable quorum. All nodes should have the
70 * We recommend a dedicated NIC for the cluster traffic, especially if
71 you use shared storage.
73 * Root password of a cluster node is required for adding nodes.
75 NOTE: It is not possible to mix {pve} 3.x and earlier with {pve} 4.X cluster
78 NOTE: While it's possible for {pve} 4.4 and {pve} 5.0 this is not supported as
79 production configuration and should only used temporarily during upgrading the
80 whole cluster from one to another major version.
82 NOTE: Running a cluster of {pve} 6.x with earlier versions is not possible. The
83 cluster protocol (corosync) between {pve} 6.x and earlier versions changed
84 fundamentally. The corosync 3 packages for {pve} 5.4 are only intended for the
85 upgrade procedure to {pve} 6.0.
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 common to reference all nodenames and their IPs in `/etc/hosts` (or
100 make their names resolvable through other means), this is not necessary for a
101 cluster to work. It may be useful however, as you can then connect from one node
102 to the other with SSH via the easier to remember node name (see also
103 xref:pvecm_corosync_addresses[Link Address Types]). Note that we always
104 recommend to reference nodes by their IP addresses in the cluster configuration.
107 [[pvecm_create_cluster]]
111 Use a unique name for your cluster. This name cannot be changed later. The
112 cluster name follows the same rules as node names.
117 Under __Datacenter -> Cluster__, click on *Create Cluster*. Enter the cluster
118 name and select a network connection from the dropdown to serve as the main
119 cluster network (Link 0). It defaults to the IP resolved via the node's
122 To add a second link as fallback, you can select the 'Advanced' checkbox and
123 choose an additional network interface (Link 1, see also
124 xref:pvecm_redundancy[Corosync Redundancy]).
126 Create via Command Line
127 ~~~~~~~~~~~~~~~~~~~~~~~
129 Login via `ssh` to the first {pve} node and run the following command:
132 hp1# pvecm create CLUSTERNAME
135 To check the state of the new cluster use:
141 Multiple Clusters In Same Network
142 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
144 It is possible to create multiple clusters in the same physical or logical
145 network. Each such cluster must have a unique name to avoid possible clashes in
146 the cluster communication stack. This also helps avoid human confusion by making
147 clusters clearly distinguishable.
149 While the bandwidth requirement of a corosync cluster is relatively low, the
150 latency of packages and the package per second (PPS) rate is the limiting
151 factor. Different clusters in the same network can compete with each other for
152 these resources, so it may still make sense to use separate physical network
153 infrastructure for bigger clusters.
155 [[pvecm_join_node_to_cluster]]
156 Adding Nodes to the Cluster
157 ---------------------------
159 CAUTION: A node that is about to be added to the cluster cannot hold any guests.
160 All existing configuration in `/etc/pve` is overwritten when joining a cluster,
161 since guest IDs could be conflicting. As a workaround create a backup of the
162 guest (`vzdump`) and restore it as a different ID after the node has been added
168 Login to the web interface on an existing cluster node. Under __Datacenter ->
169 Cluster__, click the button *Join Information* at the top. Then, click on the
170 button *Copy Information*. Alternatively, copy the string from the 'Information'
173 Next, login to the web interface on the node you want to add.
174 Under __Datacenter -> Cluster__, click on *Join Cluster*. Fill in the
175 'Information' field with the text you copied earlier.
177 For security reasons, the cluster password has to be entered manually.
179 NOTE: To enter all required data manually, you can disable the 'Assisted Join'
182 After clicking on *Join* the node will immediately be added to the cluster. You
183 might need to reload the web page and re-login with the cluster credentials.
185 Confirm that your node is visible under __Datacenter -> Cluster__.
187 Add Node via Command Line
188 ~~~~~~~~~~~~~~~~~~~~~~~~~
190 Login via `ssh` to the node you want to add.
193 hp2# pvecm add IP-ADDRESS-CLUSTER
196 For `IP-ADDRESS-CLUSTER` use the IP or hostname of an existing cluster node.
197 An IP address is recommended (see xref:pvecm_corosync_addresses[Link Address Types]).
200 To check the state of the cluster use:
206 .Cluster status after adding 4 nodes
211 Date: Mon Apr 20 12:30:13 2015
212 Quorum provider: corosync_votequorum
218 Votequorum information
219 ~~~~~~~~~~~~~~~~~~~~~~
226 Membership information
227 ~~~~~~~~~~~~~~~~~~~~~~
229 0x00000001 1 192.168.15.91
230 0x00000002 1 192.168.15.92 (local)
231 0x00000003 1 192.168.15.93
232 0x00000004 1 192.168.15.94
235 If you only want the list of all nodes use:
241 .List nodes in a cluster
245 Membership information
246 ~~~~~~~~~~~~~~~~~~~~~~
254 [[pvecm_adding_nodes_with_separated_cluster_network]]
255 Adding Nodes With Separated Cluster Network
256 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
258 When adding a node to a cluster with a separated cluster network you need to
259 use the 'link0' parameter to set the nodes address on that network:
263 pvecm add IP-ADDRESS-CLUSTER -link0 LOCAL-IP-ADDRESS-LINK0
266 If you want to use the built-in xref:pvecm_redundancy[redundancy] of the
267 kronosnet transport layer, also use the 'link1' parameter.
269 Using the GUI, you can select the correct interface from the corresponding 'Link 0'
270 and 'Link 1' fields in the *Cluster Join* dialog.
272 Remove a Cluster Node
273 ---------------------
275 CAUTION: Read carefully the procedure before proceeding, as it could
276 not be what you want or need.
278 Move all virtual machines from the node. Make sure you have no local
279 data or backups you want to keep, or save them accordingly.
280 In the following example we will remove the node hp4 from the cluster.
282 Log in to a *different* cluster node (not hp4), and issue a `pvecm nodes`
283 command to identify the node ID to remove:
288 Membership information
289 ~~~~~~~~~~~~~~~~~~~~~~
298 At this point you must power off hp4 and
299 make sure that it will not power on again (in the network) as it
302 IMPORTANT: As said above, it is critical to power off the node
303 *before* removal, and make sure that it will *never* power on again
304 (in the existing cluster network) as it is.
305 If you power on the node as it is, your cluster will be screwed up and
306 it could be difficult to restore a clean cluster state.
308 After powering off the node hp4, we can safely remove it from the cluster.
311 hp1# pvecm delnode hp4
314 If the operation succeeds no output is returned, just check the node
315 list again with `pvecm nodes` or `pvecm status`. You should see
323 Date: Mon Apr 20 12:44:28 2015
324 Quorum provider: corosync_votequorum
330 Votequorum information
331 ~~~~~~~~~~~~~~~~~~~~~~
338 Membership information
339 ~~~~~~~~~~~~~~~~~~~~~~
341 0x00000001 1 192.168.15.90 (local)
342 0x00000002 1 192.168.15.91
343 0x00000003 1 192.168.15.92
346 If, for whatever reason, you want this server to join the same cluster again,
349 * reinstall {pve} on it from scratch
351 * then join it, as explained in the previous section.
353 NOTE: After removal of the node, its SSH fingerprint will still reside in the
354 'known_hosts' of the other nodes. If you receive an SSH error after rejoining
355 a node with the same IP or hostname, run `pvecm updatecerts` once on the
356 re-added node to update its fingerprint cluster wide.
358 [[pvecm_separate_node_without_reinstall]]
359 Separate A Node Without Reinstalling
360 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
362 CAUTION: This is *not* the recommended method, proceed with caution. Use the
363 above mentioned method if you're unsure.
365 You can also separate a node from a cluster without reinstalling it from
366 scratch. But after removing the node from the cluster it will still have
367 access to the shared storages! This must be resolved before you start removing
368 the node from the cluster. A {pve} cluster cannot share the exact same
369 storage with another cluster, as storage locking doesn't work over cluster
370 boundary. Further, it may also lead to VMID conflicts.
372 Its suggested that you create a new storage where only the node which you want
373 to separate has access. This can be a new export on your NFS or a new Ceph
374 pool, to name a few examples. Its just important that the exact same storage
375 does not gets accessed by multiple clusters. After setting this storage up move
376 all data from the node and its VMs to it. Then you are ready to separate the
377 node from the cluster.
379 WARNING: Ensure all shared resources are cleanly separated! Otherwise you will
380 run into conflicts and problems.
382 First stop the corosync and the pve-cluster services on the node:
385 systemctl stop pve-cluster
386 systemctl stop corosync
389 Start the cluster filesystem again in local mode:
395 Delete the corosync configuration files:
398 rm /etc/pve/corosync.conf
402 You can now start the filesystem again as normal service:
406 systemctl start pve-cluster
409 The node is now separated from the cluster. You can deleted it from a remaining
410 node of the cluster with:
413 pvecm delnode oldnode
416 If the command failed, because the remaining node in the cluster lost quorum
417 when the now separate node exited, you may set the expected votes to 1 as a workaround:
423 And then repeat the 'pvecm delnode' command.
425 Now switch back to the separated node, here delete all remaining files left
426 from the old cluster. This ensures that the node can be added to another
427 cluster again without problems.
431 rm /var/lib/corosync/*
434 As the configuration files from the other nodes are still in the cluster
435 filesystem you may want to clean those up too. Remove simply the whole
436 directory recursive from '/etc/pve/nodes/NODENAME', but check three times that
437 you used the correct one before deleting it.
439 CAUTION: The nodes SSH keys are still in the 'authorized_key' file, this means
440 the nodes can still connect to each other with public key authentication. This
441 should be fixed by removing the respective keys from the
442 '/etc/pve/priv/authorized_keys' file.
448 {pve} use a quorum-based technique to provide a consistent state among
451 [quote, from Wikipedia, Quorum (distributed computing)]
453 A quorum is the minimum number of votes that a distributed transaction
454 has to obtain in order to be allowed to perform an operation in a
458 In case of network partitioning, state changes requires that a
459 majority of nodes are online. The cluster switches to read-only mode
462 NOTE: {pve} assigns a single vote to each node by default.
468 The cluster network is the core of a cluster. All messages sent over it have to
469 be delivered reliably to all nodes in their respective order. In {pve} this
470 part is done by corosync, an implementation of a high performance, low overhead
471 high availability development toolkit. It serves our decentralized
472 configuration file system (`pmxcfs`).
474 [[pvecm_cluster_network_requirements]]
477 This needs a reliable network with latencies under 2 milliseconds (LAN
478 performance) to work properly. The network should not be used heavily by other
479 members, ideally corosync runs on its own network. Do not use a shared network
480 for corosync and storage (except as a potential low-priority fallback in a
481 xref:pvecm_redundancy[redundant] configuration).
483 Before setting up a cluster, it is good practice to check if the network is fit
484 for that purpose. To make sure the nodes can connect to each other on the
485 cluster network, you can test the connectivity between them with the `ping`
488 If the {pve} firewall is enabled, ACCEPT rules for corosync will automatically
489 be generated - no manual action is required.
491 NOTE: Corosync used Multicast before version 3.0 (introduced in {pve} 6.0).
492 Modern versions rely on https://kronosnet.org/[Kronosnet] for cluster
493 communication, which, for now, only supports regular UDP unicast.
495 CAUTION: You can still enable Multicast or legacy unicast by setting your
496 transport to `udp` or `udpu` in your xref:pvecm_edit_corosync_conf[corosync.conf],
497 but keep in mind that this will disable all cryptography and redundancy support.
498 This is therefore not recommended.
500 Separate Cluster Network
501 ~~~~~~~~~~~~~~~~~~~~~~~~
503 When creating a cluster without any parameters the corosync cluster network is
504 generally shared with the Web UI and the VMs and their traffic. Depending on
505 your setup, even storage traffic may get sent over the same network. Its
506 recommended to change that, as corosync is a time critical real time
509 Setting Up A New Network
510 ^^^^^^^^^^^^^^^^^^^^^^^^
512 First you have to set up a new network interface. It should be on a physically
513 separate network. Ensure that your network fulfills the
514 xref:pvecm_cluster_network_requirements[cluster network requirements].
516 Separate On Cluster Creation
517 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
519 This is possible via the 'linkX' parameters of the 'pvecm create'
520 command used for creating a new cluster.
522 If you have set up an additional NIC with a static address on 10.10.10.1/25,
523 and want to send and receive all cluster communication over this interface,
528 pvecm create test --link0 10.10.10.1
531 To check if everything is working properly execute:
534 systemctl status corosync
537 Afterwards, proceed as described above to
538 xref:pvecm_adding_nodes_with_separated_cluster_network[add nodes with a separated cluster network].
540 [[pvecm_separate_cluster_net_after_creation]]
541 Separate After Cluster Creation
542 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
544 You can do this if you have already created a cluster and want to switch
545 its communication to another network, without rebuilding the whole cluster.
546 This change may lead to short durations of quorum loss in the cluster, as nodes
547 have to restart corosync and come up one after the other on the new network.
549 Check how to xref:pvecm_edit_corosync_conf[edit the corosync.conf file] first.
550 Then, open it and you should see a file similar to:
584 provider: corosync_votequorum
588 cluster_name: testcluster
600 NOTE: `ringX_addr` actually specifies a corosync *link address*, the name "ring"
601 is a remnant of older corosync versions that is kept for backwards
604 The first thing you want to do is add the 'name' properties in the node entries
605 if you do not see them already. Those *must* match the node name.
607 Then replace all addresses from the 'ring0_addr' properties of all nodes with
608 the new addresses. You may use plain IP addresses or hostnames here. If you use
609 hostnames ensure that they are resolvable from all nodes. (see also
610 xref:pvecm_corosync_addresses[Link Address Types])
612 In this example, we want to switch the cluster communication to the
613 10.10.10.1/25 network. So we replace all 'ring0_addr' respectively.
615 NOTE: The exact same procedure can be used to change other 'ringX_addr' values
616 as well, although we recommend to not change multiple addresses at once, to make
617 it easier to recover if something goes wrong.
619 After we increase the 'config_version' property, the new configuration file
634 ring0_addr: 10.10.10.2
641 ring0_addr: 10.10.10.3
648 ring0_addr: 10.10.10.1
654 provider: corosync_votequorum
658 cluster_name: testcluster
670 Then, after a final check if all changed information is correct, we save it and
671 once again follow the xref:pvecm_edit_corosync_conf[edit corosync.conf file]
672 section to bring it into effect.
674 The changes will be applied live, so restarting corosync is not strictly
675 necessary. If you changed other settings as well, or notice corosync
676 complaining, you can optionally trigger a restart.
678 On a single node execute:
682 systemctl restart corosync
685 Now check if everything is fine:
689 systemctl status corosync
692 If corosync runs again correct restart corosync also on all other nodes.
693 They will then join the cluster membership one by one on the new network.
695 [[pvecm_corosync_addresses]]
699 A corosync link address (for backwards compatibility denoted by 'ringX_addr' in
700 `corosync.conf`) can be specified in two ways:
702 * **IPv4/v6 addresses** will be used directly. They are recommended, since they
703 are static and usually not changed carelessly.
705 * **Hostnames** will be resolved using `getaddrinfo`, which means that per
706 default, IPv6 addresses will be used first, if available (see also
707 `man gai.conf`). Keep this in mind, especially when upgrading an existing
710 CAUTION: Hostnames should be used with care, since the address they
711 resolve to can be changed without touching corosync or the node it runs on -
712 which may lead to a situation where an address is changed without thinking
713 about implications for corosync.
715 A seperate, static hostname specifically for corosync is recommended, if
716 hostnames are preferred. Also, make sure that every node in the cluster can
717 resolve all hostnames correctly.
719 Since {pve} 5.1, while supported, hostnames will be resolved at the time of
720 entry. Only the resolved IP is then saved to the configuration.
722 Nodes that joined the cluster on earlier versions likely still use their
723 unresolved hostname in `corosync.conf`. It might be a good idea to replace
724 them with IPs or a seperate hostname, as mentioned above.
731 Corosync supports redundant networking via its integrated kronosnet layer by
732 default (it is not supported on the legacy udp/udpu transports). It can be
733 enabled by specifying more than one link address, either via the '--linkX'
734 parameters of `pvecm`, in the GUI as **Link 1** (while creating a cluster or
735 adding a new node) or by specifying more than one 'ringX_addr' in
738 NOTE: To provide useful failover, every link should be on its own
739 physical network connection.
741 Links are used according to a priority setting. You can configure this priority
742 by setting 'knet_link_priority' in the corresponding interface section in
743 `corosync.conf`, or, preferrably, using the 'priority' parameter when creating
744 your cluster with `pvecm`:
747 # pvecm create CLUSTERNAME --link0 10.10.10.1,priority=20 --link1 10.20.20.1,priority=15
750 This would cause 'link1' to be used first, since it has the lower priority.
752 If no priorities are configured manually (or two links have the same priority),
753 links will be used in order of their number, with the lower number having higher
756 Even if all links are working, only the one with the highest priority will see
757 corosync traffic. Link priorities cannot be mixed, i.e. links with different
758 priorities will not be able to communicate with each other.
760 Since lower priority links will not see traffic unless all higher priorities
761 have failed, it becomes a useful strategy to specify even networks used for
762 other tasks (VMs, storage, etc...) as low-priority links. If worst comes to
763 worst, a higher-latency or more congested connection might be better than no
766 Adding Redundant Links To An Existing Cluster
767 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
769 To add a new link to a running configuration, first check how to
770 xref:pvecm_edit_corosync_conf[edit the corosync.conf file].
772 Then, add a new 'ringX_addr' to every node in the `nodelist` section. Make
773 sure that your 'X' is the same for every node you add it to, and that it is
774 unique for each node.
776 Lastly, add a new 'interface', as shown below, to your `totem`
777 section, replacing 'X' with your link number chosen above.
779 Assuming you added a link with number 1, the new configuration file could look
794 ring0_addr: 10.10.10.2
795 ring1_addr: 10.20.20.2
802 ring0_addr: 10.10.10.3
803 ring1_addr: 10.20.20.3
810 ring0_addr: 10.10.10.1
811 ring1_addr: 10.20.20.1
817 provider: corosync_votequorum
821 cluster_name: testcluster
835 The new link will be enabled as soon as you follow the last steps to
836 xref:pvecm_edit_corosync_conf[edit the corosync.conf file]. A restart should not
837 be necessary. You can check that corosync loaded the new link using:
840 journalctl -b -u corosync
843 It might be a good idea to test the new link by temporarily disconnecting the
844 old link on one node and making sure that its status remains online while
851 If you see a healthy cluster state, it means that your new link is being used.
854 Corosync External Vote Support
855 ------------------------------
857 This section describes a way to deploy an external voter in a {pve} cluster.
858 When configured, the cluster can sustain more node failures without
859 violating safety properties of the cluster communication.
861 For this to work there are two services involved:
863 * a so called qdevice daemon which runs on each {pve} node
865 * an external vote daemon which runs on an independent server.
867 As a result you can achieve higher availability even in smaller setups (for
870 QDevice Technical Overview
871 ~~~~~~~~~~~~~~~~~~~~~~~~~~
873 The Corosync Quroum Device (QDevice) is a daemon which runs on each cluster
874 node. It provides a configured number of votes to the clusters quorum
875 subsystem based on an external running third-party arbitrator's decision.
876 Its primary use is to allow a cluster to sustain more node failures than
877 standard quorum rules allow. This can be done safely as the external device
878 can see all nodes and thus choose only one set of nodes to give its vote.
879 This will only be done if said set of nodes can have quorum (again) when
880 receiving the third-party vote.
882 Currently only 'QDevice Net' is supported as a third-party arbitrator. It is
883 a daemon which provides a vote to a cluster partition if it can reach the
884 partition members over the network. It will give only votes to one partition
885 of a cluster at any time.
886 It's designed to support multiple clusters and is almost configuration and
887 state free. New clusters are handled dynamically and no configuration file
888 is needed on the host running a QDevice.
890 The external host has the only requirement that it needs network access to the
891 cluster and a corosync-qnetd package available. We provide such a package
892 for Debian based hosts, other Linux distributions should also have a package
893 available through their respective package manager.
895 NOTE: In contrast to corosync itself, a QDevice connects to the cluster over
896 TCP/IP. The daemon may even run outside of the clusters LAN and can have longer
902 We support QDevices for clusters with an even number of nodes and recommend
903 it for 2 node clusters, if they should provide higher availability.
904 For clusters with an odd node count we discourage the use of QDevices
905 currently. The reason for this, is the difference of the votes the QDevice
906 provides for each cluster type. Even numbered clusters get single additional
907 vote, with this we can only increase availability, i.e. if the QDevice
908 itself fails we are in the same situation as with no QDevice at all.
910 Now, with an odd numbered cluster size the QDevice provides '(N-1)' votes --
911 where 'N' corresponds to the cluster node count. This difference makes
912 sense, if we had only one additional vote the cluster can get into a split
914 This algorithm would allow that all nodes but one (and naturally the
915 QDevice itself) could fail.
916 There are two drawbacks with this:
918 * If the QNet daemon itself fails, no other node may fail or the cluster
919 immediately loses quorum. For example, in a cluster with 15 nodes 7
920 could fail before the cluster becomes inquorate. But, if a QDevice is
921 configured here and said QDevice fails itself **no single node** of
922 the 15 may fail. The QDevice acts almost as a single point of failure in
925 * The fact that all but one node plus QDevice may fail sound promising at
926 first, but this may result in a mass recovery of HA services that would
927 overload the single node left. Also ceph server will stop to provide
928 services after only '((N-1)/2)' nodes are online.
930 If you understand the drawbacks and implications you can decide yourself if
931 you should use this technology in an odd numbered cluster setup.
936 We recommend to run any daemon which provides votes to corosync-qdevice as an
937 unprivileged user. {pve} and Debian provides a package which is already
939 The traffic between the daemon and the cluster must be encrypted to ensure a
940 safe and secure QDevice integration in {pve}.
942 First install the 'corosync-qnetd' package on your external server and
943 the 'corosync-qdevice' package on all cluster nodes.
945 After that, ensure that all your nodes on the cluster are online.
947 You can now easily set up your QDevice by running the following command on one
951 pve# pvecm qdevice setup <QDEVICE-IP>
954 The SSH key from the cluster will be automatically copied to the QDevice. You
955 might need to enter an SSH password during this step.
957 After you enter the password and all the steps are successfully completed, you
958 will see "Done". You can check the status now:
965 Votequorum information
966 ~~~~~~~~~~~~~~~~~~~~~
971 Flags: Quorate Qdevice
973 Membership information
974 ~~~~~~~~~~~~~~~~~~~~~~
975 Nodeid Votes Qdevice Name
976 0x00000001 1 A,V,NMW 192.168.22.180 (local)
977 0x00000002 1 A,V,NMW 192.168.22.181
982 which means the QDevice is set up.
984 Frequently Asked Questions
985 ~~~~~~~~~~~~~~~~~~~~~~~~~~
990 In case of a tie, where two same-sized cluster partitions cannot see each other
991 but the QDevice, the QDevice chooses randomly one of those partitions and
992 provides a vote to it.
994 Possible Negative Implications
995 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
997 For clusters with an even node count there are no negative implications when
998 setting up a QDevice. If it fails to work, you are as good as without QDevice at
1001 Adding/Deleting Nodes After QDevice Setup
1002 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1004 If you want to add a new node or remove an existing one from a cluster with a
1005 QDevice setup, you need to remove the QDevice first. After that, you can add or
1006 remove nodes normally. Once you have a cluster with an even node count again,
1007 you can set up the QDevice again as described above.
1009 Removing the QDevice
1010 ^^^^^^^^^^^^^^^^^^^^
1012 If you used the official `pvecm` tool to add the QDevice, you can remove it
1013 trivially by running:
1016 pve# pvecm qdevice remove
1021 //There is still stuff to add here
1024 Corosync Configuration
1025 ----------------------
1027 The `/etc/pve/corosync.conf` file plays a central role in a {pve} cluster. It
1028 controls the cluster membership and its network.
1029 For further information about it, check the corosync.conf man page:
1035 For node membership you should always use the `pvecm` tool provided by {pve}.
1036 You may have to edit the configuration file manually for other changes.
1037 Here are a few best practice tips for doing this.
1039 [[pvecm_edit_corosync_conf]]
1043 Editing the corosync.conf file is not always very straightforward. There are
1044 two on each cluster node, one in `/etc/pve/corosync.conf` and the other in
1045 `/etc/corosync/corosync.conf`. Editing the one in our cluster file system will
1046 propagate the changes to the local one, but not vice versa.
1048 The configuration will get updated automatically as soon as the file changes.
1049 This means changes which can be integrated in a running corosync will take
1050 effect immediately. So you should always make a copy and edit that instead, to
1051 avoid triggering some unwanted changes by an in-between safe.
1055 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
1058 Then open the config file with your favorite editor, `nano` and `vim.tiny` are
1059 preinstalled on any {pve} node for example.
1061 NOTE: Always increment the 'config_version' number on configuration changes,
1062 omitting this can lead to problems.
1064 After making the necessary changes create another copy of the current working
1065 configuration file. This serves as a backup if the new configuration fails to
1066 apply or makes problems in other ways.
1070 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak
1073 Then move the new configuration file over the old one:
1076 mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
1079 You may check with the commands
1082 systemctl status corosync
1083 journalctl -b -u corosync
1086 If the change could be applied automatically. If not you may have to restart the
1087 corosync service via:
1090 systemctl restart corosync
1093 On errors check the troubleshooting section below.
1098 Issue: 'quorum.expected_votes must be configured'
1099 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1101 When corosync starts to fail and you get the following message in the system log:
1105 corosync[1647]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
1106 corosync[1647]: [SERV ] Service engine 'corosync_quorum' failed to load for reason
1107 'configuration error: nodelist or quorum.expected_votes must be configured!'
1111 It means that the hostname you set for corosync 'ringX_addr' in the
1112 configuration could not be resolved.
1114 Write Configuration When Not Quorate
1115 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1117 If you need to change '/etc/pve/corosync.conf' on an node with no quorum, and you
1118 know what you do, use:
1124 This sets the expected vote count to 1 and makes the cluster quorate. You can
1125 now fix your configuration, or revert it back to the last working backup.
1127 This is not enough if corosync cannot start anymore. Here it is best to edit the
1128 local copy of the corosync configuration in '/etc/corosync/corosync.conf' so
1129 that corosync can start again. Ensure that on all nodes this configuration has
1130 the same content to avoid split brains. If you are not sure what went wrong
1131 it's best to ask the Proxmox Community to help you.
1134 [[pvecm_corosync_conf_glossary]]
1135 Corosync Configuration Glossary
1136 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1139 This names the different link addresses for the kronosnet connections between
1146 It is obvious that a cluster is not quorate when all nodes are
1147 offline. This is a common case after a power failure.
1149 NOTE: It is always a good idea to use an uninterruptible power supply
1150 (``UPS'', also called ``battery backup'') to avoid this state, especially if
1153 On node startup, the `pve-guests` service is started and waits for
1154 quorum. Once quorate, it starts all guests which have the `onboot`
1157 When you turn on nodes, or when power comes back after power failure,
1158 it is likely that some nodes boots faster than others. Please keep in
1159 mind that guest startup is delayed until you reach quorum.
1165 Migrating virtual guests to other nodes is a useful feature in a
1166 cluster. There are settings to control the behavior of such
1167 migrations. This can be done via the configuration file
1168 `datacenter.cfg` or for a specific migration via API or command line
1171 It makes a difference if a Guest is online or offline, or if it has
1172 local resources (like a local disk).
1174 For Details about Virtual Machine Migration see the
1175 xref:qm_migration[QEMU/KVM Migration Chapter].
1177 For Details about Container Migration see the
1178 xref:pct_migration[Container Migration Chapter].
1183 The migration type defines if the migration data should be sent over an
1184 encrypted (`secure`) channel or an unencrypted (`insecure`) one.
1185 Setting the migration type to insecure means that the RAM content of a
1186 virtual guest gets also transferred unencrypted, which can lead to
1187 information disclosure of critical data from inside the guest (for
1188 example passwords or encryption keys).
1190 Therefore, we strongly recommend using the secure channel if you do
1191 not have full control over the network and can not guarantee that no
1192 one is eavesdropping on it.
1194 NOTE: Storage migration does not follow this setting. Currently, it
1195 always sends the storage content over a secure channel.
1197 Encryption requires a lot of computing power, so this setting is often
1198 changed to "unsafe" to achieve better performance. The impact on
1199 modern systems is lower because they implement AES encryption in
1200 hardware. The performance impact is particularly evident in fast
1201 networks where you can transfer 10 Gbps or more.
1206 By default, {pve} uses the network in which cluster communication
1207 takes place to send the migration traffic. This is not optimal because
1208 sensitive cluster traffic can be disrupted and this network may not
1209 have the best bandwidth available on the node.
1211 Setting the migration network parameter allows the use of a dedicated
1212 network for the entire migration traffic. In addition to the memory,
1213 this also affects the storage traffic for offline migrations.
1215 The migration network is set as a network in the CIDR notation. This
1216 has the advantage that you do not have to set individual IP addresses
1217 for each node. {pve} can determine the real address on the
1218 destination node from the network specified in the CIDR form. To
1219 enable this, the network must be specified so that each node has one,
1220 but only one IP in the respective network.
1225 We assume that we have a three-node setup with three separate
1226 networks. One for public communication with the Internet, one for
1227 cluster communication and a very fast one, which we want to use as a
1228 dedicated network for migration.
1230 A network configuration for such a setup might look as follows:
1233 iface eno1 inet manual
1237 iface vmbr0 inet static
1239 netmask 255.255.250.0
1247 iface eno2 inet static
1249 netmask 255.255.255.0
1253 iface eno3 inet static
1255 netmask 255.255.255.0
1258 Here, we will use the network 10.1.2.0/24 as a migration network. For
1259 a single migration, you can do this using the `migration_network`
1260 parameter of the command line tool:
1263 # qm migrate 106 tre --online --migration_network 10.1.2.0/24
1266 To configure this as the default network for all migrations in the
1267 cluster, set the `migration` property of the `/etc/pve/datacenter.cfg`
1271 # use dedicated migration network
1272 migration: secure,network=10.1.2.0/24
1275 NOTE: The migration type must always be set when the migration network
1276 gets set in `/etc/pve/datacenter.cfg`.
1280 include::pve-copyright.adoc[]