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 to mix {pve} 4.4 and {pve} 5.0 nodes, doing so is
79 not supported as production configuration and should only used temporarily
80 during upgrading the 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 While it's common to reference all nodenames and their IPs in `/etc/hosts` (or
96 make their names resolvable through other means), this is not necessary for a
97 cluster to work. It may be useful however, as you can then connect from one node
98 to the other with SSH via the easier to remember node name (see also
99 xref:pvecm_corosync_addresses[Link Address Types]). Note that we always
100 recommend to reference nodes by their IP addresses in the cluster configuration.
103 [[pvecm_create_cluster]]
107 You can either create a cluster on the console (login via `ssh`), or through
108 the API using the {pve} Webinterface (__Datacenter -> Cluster__).
110 NOTE: Use a unique name for your cluster. This name cannot be changed later.
111 The cluster name follows the same rules as node names.
113 [[pvecm_cluster_create_via_gui]]
117 [thumbnail="screenshot/gui-cluster-create.png"]
119 Under __Datacenter -> Cluster__, click on *Create Cluster*. Enter the cluster
120 name and select a network connection from the dropdown to serve as the main
121 cluster network (Link 0). It defaults to the IP resolved via the node's
124 To add a second link as fallback, you can select the 'Advanced' checkbox and
125 choose an additional network interface (Link 1, see also
126 xref:pvecm_redundancy[Corosync Redundancy]).
128 NOTE: Ensure the network selected for the cluster communication is not used for
129 any high traffic loads like those of (network) storages or live-migration.
130 While the cluster network itself produces small amounts of data, it is very
131 sensitive to latency. Check out full
132 xref:pvecm_cluster_network_requirements[cluster network requirements].
134 [[pvecm_cluster_create_via_cli]]
135 Create via Command Line
136 ~~~~~~~~~~~~~~~~~~~~~~~
138 Login via `ssh` to the first {pve} node and run the following command:
141 hp1# pvecm create CLUSTERNAME
144 To check the state of the new cluster use:
150 Multiple Clusters In Same Network
151 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
153 It is possible to create multiple clusters in the same physical or logical
154 network. Each such cluster must have a unique name to avoid possible clashes in
155 the cluster communication stack. This also helps avoid human confusion by making
156 clusters clearly distinguishable.
158 While the bandwidth requirement of a corosync cluster is relatively low, the
159 latency of packages and the package per second (PPS) rate is the limiting
160 factor. Different clusters in the same network can compete with each other for
161 these resources, so it may still make sense to use separate physical network
162 infrastructure for bigger clusters.
164 [[pvecm_join_node_to_cluster]]
165 Adding Nodes to the Cluster
166 ---------------------------
168 CAUTION: A node that is about to be added to the cluster cannot hold any guests.
169 All existing configuration in `/etc/pve` is overwritten when joining a cluster,
170 since guest IDs could be conflicting. As a workaround create a backup of the
171 guest (`vzdump`) and restore it as a different ID after the node has been added
174 Join Node to Cluster via GUI
175 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
177 [thumbnail="screenshot/gui-cluster-join-information.png"]
179 Login to the web interface on an existing cluster node. Under __Datacenter ->
180 Cluster__, click the button *Join Information* at the top. Then, click on the
181 button *Copy Information*. Alternatively, copy the string from the 'Information'
184 [thumbnail="screenshot/gui-cluster-join.png"]
186 Next, login to the web interface on the node you want to add.
187 Under __Datacenter -> Cluster__, click on *Join Cluster*. Fill in the
188 'Information' field with the 'Join Information' text you copied earlier.
189 Most settings required for joining the cluster will be filled out
190 automatically. For security reasons, the cluster password has to be entered
193 NOTE: To enter all required data manually, you can disable the 'Assisted Join'
196 After clicking the *Join* button, the cluster join process will start
197 immediately. After the node joined the cluster its current node certificate
198 will be replaced by one signed from the cluster certificate authority (CA),
199 that means the current session will stop to work after a few seconds. You might
200 then need to force-reload the webinterface and re-login with the cluster
203 Now your node should be visible under __Datacenter -> Cluster__.
205 Join Node to Cluster via Command Line
206 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
208 Login via `ssh` to the node you want to join into an existing cluster.
211 hp2# pvecm add IP-ADDRESS-CLUSTER
214 For `IP-ADDRESS-CLUSTER` use the IP or hostname of an existing cluster node.
215 An IP address is recommended (see xref:pvecm_corosync_addresses[Link Address Types]).
218 To check the state of the cluster use:
224 .Cluster status after adding 4 nodes
229 Date: Mon Apr 20 12:30:13 2015
230 Quorum provider: corosync_votequorum
236 Votequorum information
237 ~~~~~~~~~~~~~~~~~~~~~~
244 Membership information
245 ~~~~~~~~~~~~~~~~~~~~~~
247 0x00000001 1 192.168.15.91
248 0x00000002 1 192.168.15.92 (local)
249 0x00000003 1 192.168.15.93
250 0x00000004 1 192.168.15.94
253 If you only want the list of all nodes use:
259 .List nodes in a cluster
263 Membership information
264 ~~~~~~~~~~~~~~~~~~~~~~
272 [[pvecm_adding_nodes_with_separated_cluster_network]]
273 Adding Nodes With Separated Cluster Network
274 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
276 When adding a node to a cluster with a separated cluster network you need to
277 use the 'link0' parameter to set the nodes address on that network:
281 pvecm add IP-ADDRESS-CLUSTER -link0 LOCAL-IP-ADDRESS-LINK0
284 If you want to use the built-in xref:pvecm_redundancy[redundancy] of the
285 kronosnet transport layer, also use the 'link1' parameter.
287 Using the GUI, you can select the correct interface from the corresponding 'Link 0'
288 and 'Link 1' fields in the *Cluster Join* dialog.
290 Remove a Cluster Node
291 ---------------------
293 CAUTION: Read carefully the procedure before proceeding, as it could
294 not be what you want or need.
296 Move all virtual machines from the node. Make sure you have no local
297 data or backups you want to keep, or save them accordingly.
298 In the following example we will remove the node hp4 from the cluster.
300 Log in to a *different* cluster node (not hp4), and issue a `pvecm nodes`
301 command to identify the node ID to remove:
306 Membership information
307 ~~~~~~~~~~~~~~~~~~~~~~
316 At this point you must power off hp4 and
317 make sure that it will not power on again (in the network) as it
320 IMPORTANT: As said above, it is critical to power off the node
321 *before* removal, and make sure that it will *never* power on again
322 (in the existing cluster network) as it is.
323 If you power on the node as it is, your cluster will be screwed up and
324 it could be difficult to restore a clean cluster state.
326 After powering off the node hp4, we can safely remove it from the cluster.
329 hp1# pvecm delnode hp4
333 Use `pvecm nodes` or `pvecm status` to check the node list again. It should
341 Date: Mon Apr 20 12:44:28 2015
342 Quorum provider: corosync_votequorum
348 Votequorum information
349 ~~~~~~~~~~~~~~~~~~~~~~
356 Membership information
357 ~~~~~~~~~~~~~~~~~~~~~~
359 0x00000001 1 192.168.15.90 (local)
360 0x00000002 1 192.168.15.91
361 0x00000003 1 192.168.15.92
364 If, for whatever reason, you want this server to join the same cluster again,
367 * reinstall {pve} on it from scratch
369 * then join it, as explained in the previous section.
371 NOTE: After removal of the node, its SSH fingerprint will still reside in the
372 'known_hosts' of the other nodes. If you receive an SSH error after rejoining
373 a node with the same IP or hostname, run `pvecm updatecerts` once on the
374 re-added node to update its fingerprint cluster wide.
376 [[pvecm_separate_node_without_reinstall]]
377 Separate A Node Without Reinstalling
378 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
380 CAUTION: This is *not* the recommended method, proceed with caution. Use the
381 above mentioned method if you're unsure.
383 You can also separate a node from a cluster without reinstalling it from
384 scratch. But after removing the node from the cluster it will still have
385 access to the shared storages! This must be resolved before you start removing
386 the node from the cluster. A {pve} cluster cannot share the exact same
387 storage with another cluster, as storage locking doesn't work over cluster
388 boundary. Further, it may also lead to VMID conflicts.
390 Its suggested that you create a new storage where only the node which you want
391 to separate has access. This can be a new export on your NFS or a new Ceph
392 pool, to name a few examples. Its just important that the exact same storage
393 does not gets accessed by multiple clusters. After setting this storage up move
394 all data from the node and its VMs to it. Then you are ready to separate the
395 node from the cluster.
397 WARNING: Ensure all shared resources are cleanly separated! Otherwise you will
398 run into conflicts and problems.
400 First, stop the corosync and the pve-cluster services on the node:
403 systemctl stop pve-cluster
404 systemctl stop corosync
407 Start the cluster filesystem again in local mode:
413 Delete the corosync configuration files:
416 rm /etc/pve/corosync.conf
417 rm -r /etc/corosync/*
420 You can now start the filesystem again as normal service:
424 systemctl start pve-cluster
427 The node is now separated from the cluster. You can deleted it from a remaining
428 node of the cluster with:
431 pvecm delnode oldnode
434 If the command failed, because the remaining node in the cluster lost quorum
435 when the now separate node exited, you may set the expected votes to 1 as a workaround:
441 And then repeat the 'pvecm delnode' command.
443 Now switch back to the separated node, here delete all remaining files left
444 from the old cluster. This ensures that the node can be added to another
445 cluster again without problems.
449 rm /var/lib/corosync/*
452 As the configuration files from the other nodes are still in the cluster
453 filesystem you may want to clean those up too. Remove simply the whole
454 directory recursive from '/etc/pve/nodes/NODENAME', but check three times that
455 you used the correct one before deleting it.
457 CAUTION: The nodes SSH keys are still in the 'authorized_key' file, this means
458 the nodes can still connect to each other with public key authentication. This
459 should be fixed by removing the respective keys from the
460 '/etc/pve/priv/authorized_keys' file.
466 {pve} use a quorum-based technique to provide a consistent state among
469 [quote, from Wikipedia, Quorum (distributed computing)]
471 A quorum is the minimum number of votes that a distributed transaction
472 has to obtain in order to be allowed to perform an operation in a
476 In case of network partitioning, state changes requires that a
477 majority of nodes are online. The cluster switches to read-only mode
480 NOTE: {pve} assigns a single vote to each node by default.
486 The cluster network is the core of a cluster. All messages sent over it have to
487 be delivered reliably to all nodes in their respective order. In {pve} this
488 part is done by corosync, an implementation of a high performance, low overhead
489 high availability development toolkit. It serves our decentralized
490 configuration file system (`pmxcfs`).
492 [[pvecm_cluster_network_requirements]]
495 This needs a reliable network with latencies under 2 milliseconds (LAN
496 performance) to work properly. The network should not be used heavily by other
497 members, ideally corosync runs on its own network. Do not use a shared network
498 for corosync and storage (except as a potential low-priority fallback in a
499 xref:pvecm_redundancy[redundant] configuration).
501 Before setting up a cluster, it is good practice to check if the network is fit
502 for that purpose. To make sure the nodes can connect to each other on the
503 cluster network, you can test the connectivity between them with the `ping`
506 If the {pve} firewall is enabled, ACCEPT rules for corosync will automatically
507 be generated - no manual action is required.
509 NOTE: Corosync used Multicast before version 3.0 (introduced in {pve} 6.0).
510 Modern versions rely on https://kronosnet.org/[Kronosnet] for cluster
511 communication, which, for now, only supports regular UDP unicast.
513 CAUTION: You can still enable Multicast or legacy unicast by setting your
514 transport to `udp` or `udpu` in your xref:pvecm_edit_corosync_conf[corosync.conf],
515 but keep in mind that this will disable all cryptography and redundancy support.
516 This is therefore not recommended.
518 Separate Cluster Network
519 ~~~~~~~~~~~~~~~~~~~~~~~~
521 When creating a cluster without any parameters the corosync cluster network is
522 generally shared with the Web UI and the VMs and their traffic. Depending on
523 your setup, even storage traffic may get sent over the same network. Its
524 recommended to change that, as corosync is a time critical real time
527 Setting Up A New Network
528 ^^^^^^^^^^^^^^^^^^^^^^^^
530 First, you have to set up a new network interface. It should be on a physically
531 separate network. Ensure that your network fulfills the
532 xref:pvecm_cluster_network_requirements[cluster network requirements].
534 Separate On Cluster Creation
535 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
537 This is possible via the 'linkX' parameters of the 'pvecm create'
538 command used for creating a new cluster.
540 If you have set up an additional NIC with a static address on 10.10.10.1/25,
541 and want to send and receive all cluster communication over this interface,
546 pvecm create test --link0 10.10.10.1
549 To check if everything is working properly execute:
552 systemctl status corosync
555 Afterwards, proceed as described above to
556 xref:pvecm_adding_nodes_with_separated_cluster_network[add nodes with a separated cluster network].
558 [[pvecm_separate_cluster_net_after_creation]]
559 Separate After Cluster Creation
560 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
562 You can do this if you have already created a cluster and want to switch
563 its communication to another network, without rebuilding the whole cluster.
564 This change may lead to short durations of quorum loss in the cluster, as nodes
565 have to restart corosync and come up one after the other on the new network.
567 Check how to xref:pvecm_edit_corosync_conf[edit the corosync.conf file] first.
568 Then, open it and you should see a file similar to:
602 provider: corosync_votequorum
606 cluster_name: testcluster
618 NOTE: `ringX_addr` actually specifies a corosync *link address*, the name "ring"
619 is a remnant of older corosync versions that is kept for backwards
622 The first thing you want to do is add the 'name' properties in the node entries
623 if you do not see them already. Those *must* match the node name.
625 Then replace all addresses from the 'ring0_addr' properties of all nodes with
626 the new addresses. You may use plain IP addresses or hostnames here. If you use
627 hostnames ensure that they are resolvable from all nodes. (see also
628 xref:pvecm_corosync_addresses[Link Address Types])
630 In this example, we want to switch the cluster communication to the
631 10.10.10.1/25 network. So we replace all 'ring0_addr' respectively.
633 NOTE: The exact same procedure can be used to change other 'ringX_addr' values
634 as well, although we recommend to not change multiple addresses at once, to make
635 it easier to recover if something goes wrong.
637 After we increase the 'config_version' property, the new configuration file
652 ring0_addr: 10.10.10.2
659 ring0_addr: 10.10.10.3
666 ring0_addr: 10.10.10.1
672 provider: corosync_votequorum
676 cluster_name: testcluster
688 Then, after a final check if all changed information is correct, we save it and
689 once again follow the xref:pvecm_edit_corosync_conf[edit corosync.conf file]
690 section to bring it into effect.
692 The changes will be applied live, so restarting corosync is not strictly
693 necessary. If you changed other settings as well, or notice corosync
694 complaining, you can optionally trigger a restart.
696 On a single node execute:
700 systemctl restart corosync
703 Now check if everything is fine:
707 systemctl status corosync
710 If corosync runs again correct restart corosync also on all other nodes.
711 They will then join the cluster membership one by one on the new network.
713 [[pvecm_corosync_addresses]]
717 A corosync link address (for backwards compatibility denoted by 'ringX_addr' in
718 `corosync.conf`) can be specified in two ways:
720 * **IPv4/v6 addresses** will be used directly. They are recommended, since they
721 are static and usually not changed carelessly.
723 * **Hostnames** will be resolved using `getaddrinfo`, which means that per
724 default, IPv6 addresses will be used first, if available (see also
725 `man gai.conf`). Keep this in mind, especially when upgrading an existing
728 CAUTION: Hostnames should be used with care, since the address they
729 resolve to can be changed without touching corosync or the node it runs on -
730 which may lead to a situation where an address is changed without thinking
731 about implications for corosync.
733 A separate, static hostname specifically for corosync is recommended, if
734 hostnames are preferred. Also, make sure that every node in the cluster can
735 resolve all hostnames correctly.
737 Since {pve} 5.1, while supported, hostnames will be resolved at the time of
738 entry. Only the resolved IP is then saved to the configuration.
740 Nodes that joined the cluster on earlier versions likely still use their
741 unresolved hostname in `corosync.conf`. It might be a good idea to replace
742 them with IPs or a separate hostname, as mentioned above.
749 Corosync supports redundant networking via its integrated kronosnet layer by
750 default (it is not supported on the legacy udp/udpu transports). It can be
751 enabled by specifying more than one link address, either via the '--linkX'
752 parameters of `pvecm`, in the GUI as **Link 1** (while creating a cluster or
753 adding a new node) or by specifying more than one 'ringX_addr' in
756 NOTE: To provide useful failover, every link should be on its own
757 physical network connection.
759 Links are used according to a priority setting. You can configure this priority
760 by setting 'knet_link_priority' in the corresponding interface section in
761 `corosync.conf`, or, preferably, using the 'priority' parameter when creating
762 your cluster with `pvecm`:
765 # pvecm create CLUSTERNAME --link0 10.10.10.1,priority=15 --link1 10.20.20.1,priority=20
768 This would cause 'link1' to be used first, since it has the higher priority.
770 If no priorities are configured manually (or two links have the same priority),
771 links will be used in order of their number, with the lower number having higher
774 Even if all links are working, only the one with the highest priority will see
775 corosync traffic. Link priorities cannot be mixed, i.e. links with different
776 priorities will not be able to communicate with each other.
778 Since lower priority links will not see traffic unless all higher priorities
779 have failed, it becomes a useful strategy to specify even networks used for
780 other tasks (VMs, storage, etc...) as low-priority links. If worst comes to
781 worst, a higher-latency or more congested connection might be better than no
784 Adding Redundant Links To An Existing Cluster
785 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
787 To add a new link to a running configuration, first check how to
788 xref:pvecm_edit_corosync_conf[edit the corosync.conf file].
790 Then, add a new 'ringX_addr' to every node in the `nodelist` section. Make
791 sure that your 'X' is the same for every node you add it to, and that it is
792 unique for each node.
794 Lastly, add a new 'interface', as shown below, to your `totem`
795 section, replacing 'X' with your link number chosen above.
797 Assuming you added a link with number 1, the new configuration file could look
812 ring0_addr: 10.10.10.2
813 ring1_addr: 10.20.20.2
820 ring0_addr: 10.10.10.3
821 ring1_addr: 10.20.20.3
828 ring0_addr: 10.10.10.1
829 ring1_addr: 10.20.20.1
835 provider: corosync_votequorum
839 cluster_name: testcluster
853 The new link will be enabled as soon as you follow the last steps to
854 xref:pvecm_edit_corosync_conf[edit the corosync.conf file]. A restart should not
855 be necessary. You can check that corosync loaded the new link using:
858 journalctl -b -u corosync
861 It might be a good idea to test the new link by temporarily disconnecting the
862 old link on one node and making sure that its status remains online while
869 If you see a healthy cluster state, it means that your new link is being used.
872 Role of SSH in {PVE} Clusters
873 -----------------------------
875 {PVE} utilizes SSH tunnels for various features.
877 * Proxying console/shell sessions (node and guests)
879 When using the shell for node B while being connected to node A, connects to a
880 terminal proxy on node A, which is in turn connected to the login shell on node
881 B via a non-interactive SSH tunnel.
883 * VM and CT memory and local-storage migration in 'secure' mode.
885 During the migration one or more SSH tunnel(s) are established between the
886 source and target nodes, in order to exchange migration information and
887 transfer memory and disk contents.
889 * Storage replication
891 .Pitfalls due to automatic execution of `.bashrc` and siblings
894 In case you have a custom `.bashrc`, or similar files that get executed on
895 login by the configured shell, `ssh` will automatically run it once the session
896 is established successfully. This can cause some unexpected behavior, as those
897 commands may be executed with root permissions on any above described
898 operation. That can cause possible problematic side-effects!
900 In order to avoid such complications, it's recommended to add a check in
901 `/root/.bashrc` to make sure the session is interactive, and only then run
904 You can add this snippet at the beginning of your `.bashrc` file:
907 # Early exit if not running interactively to avoid side-effects!
916 Corosync External Vote Support
917 ------------------------------
919 This section describes a way to deploy an external voter in a {pve} cluster.
920 When configured, the cluster can sustain more node failures without
921 violating safety properties of the cluster communication.
923 For this to work there are two services involved:
925 * a so called qdevice daemon which runs on each {pve} node
927 * an external vote daemon which runs on an independent server.
929 As a result you can achieve higher availability even in smaller setups (for
932 QDevice Technical Overview
933 ~~~~~~~~~~~~~~~~~~~~~~~~~~
935 The Corosync Quorum Device (QDevice) is a daemon which runs on each cluster
936 node. It provides a configured number of votes to the clusters quorum
937 subsystem based on an external running third-party arbitrator's decision.
938 Its primary use is to allow a cluster to sustain more node failures than
939 standard quorum rules allow. This can be done safely as the external device
940 can see all nodes and thus choose only one set of nodes to give its vote.
941 This will only be done if said set of nodes can have quorum (again) when
942 receiving the third-party vote.
944 Currently only 'QDevice Net' is supported as a third-party arbitrator. It is
945 a daemon which provides a vote to a cluster partition if it can reach the
946 partition members over the network. It will give only votes to one partition
947 of a cluster at any time.
948 It's designed to support multiple clusters and is almost configuration and
949 state free. New clusters are handled dynamically and no configuration file
950 is needed on the host running a QDevice.
952 The external host has the only requirement that it needs network access to the
953 cluster and a corosync-qnetd package available. We provide such a package
954 for Debian based hosts, other Linux distributions should also have a package
955 available through their respective package manager.
957 NOTE: In contrast to corosync itself, a QDevice connects to the cluster over
958 TCP/IP. The daemon may even run outside of the clusters LAN and can have longer
964 We support QDevices for clusters with an even number of nodes and recommend
965 it for 2 node clusters, if they should provide higher availability.
966 For clusters with an odd node count we discourage the use of QDevices
967 currently. The reason for this, is the difference of the votes the QDevice
968 provides for each cluster type. Even numbered clusters get single additional
969 vote, with this we can only increase availability, i.e. if the QDevice
970 itself fails we are in the same situation as with no QDevice at all.
972 Now, with an odd numbered cluster size the QDevice provides '(N-1)' votes --
973 where 'N' corresponds to the cluster node count. This difference makes
974 sense, if we had only one additional vote the cluster can get into a split
976 This algorithm would allow that all nodes but one (and naturally the
977 QDevice itself) could fail.
978 There are two drawbacks with this:
980 * If the QNet daemon itself fails, no other node may fail or the cluster
981 immediately loses quorum. For example, in a cluster with 15 nodes 7
982 could fail before the cluster becomes inquorate. But, if a QDevice is
983 configured here and said QDevice fails itself **no single node** of
984 the 15 may fail. The QDevice acts almost as a single point of failure in
987 * The fact that all but one node plus QDevice may fail sound promising at
988 first, but this may result in a mass recovery of HA services that would
989 overload the single node left. Also ceph server will stop to provide
990 services after only '((N-1)/2)' nodes are online.
992 If you understand the drawbacks and implications you can decide yourself if
993 you should use this technology in an odd numbered cluster setup.
998 We recommend to run any daemon which provides votes to corosync-qdevice as an
999 unprivileged user. {pve} and Debian provide a package which is already
1000 configured to do so.
1001 The traffic between the daemon and the cluster must be encrypted to ensure a
1002 safe and secure QDevice integration in {pve}.
1004 First, install the 'corosync-qnetd' package on your external server
1007 external# apt install corosync-qnetd
1010 and the 'corosync-qdevice' package on all cluster nodes
1013 pve# apt install corosync-qdevice
1016 After that, ensure that all your nodes on the cluster are online.
1018 You can now easily set up your QDevice by running the following command on one
1022 pve# pvecm qdevice setup <QDEVICE-IP>
1025 The SSH key from the cluster will be automatically copied to the QDevice.
1027 NOTE: Make sure that the SSH configuration on your external server allows root
1028 login via password, if you are asked for a password during this step.
1030 After you enter the password and all the steps are successfully completed, you
1031 will see "Done". You can check the status now:
1038 Votequorum information
1039 ~~~~~~~~~~~~~~~~~~~~~
1044 Flags: Quorate Qdevice
1046 Membership information
1047 ~~~~~~~~~~~~~~~~~~~~~~
1048 Nodeid Votes Qdevice Name
1049 0x00000001 1 A,V,NMW 192.168.22.180 (local)
1050 0x00000002 1 A,V,NMW 192.168.22.181
1051 0x00000000 1 Qdevice
1055 which means the QDevice is set up.
1057 Frequently Asked Questions
1058 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1063 In case of a tie, where two same-sized cluster partitions cannot see each other
1064 but the QDevice, the QDevice chooses randomly one of those partitions and
1065 provides a vote to it.
1067 Possible Negative Implications
1068 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1070 For clusters with an even node count there are no negative implications when
1071 setting up a QDevice. If it fails to work, you are as good as without QDevice at
1074 Adding/Deleting Nodes After QDevice Setup
1075 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1077 If you want to add a new node or remove an existing one from a cluster with a
1078 QDevice setup, you need to remove the QDevice first. After that, you can add or
1079 remove nodes normally. Once you have a cluster with an even node count again,
1080 you can set up the QDevice again as described above.
1082 Removing the QDevice
1083 ^^^^^^^^^^^^^^^^^^^^
1085 If you used the official `pvecm` tool to add the QDevice, you can remove it
1086 trivially by running:
1089 pve# pvecm qdevice remove
1094 //There is still stuff to add here
1097 Corosync Configuration
1098 ----------------------
1100 The `/etc/pve/corosync.conf` file plays a central role in a {pve} cluster. It
1101 controls the cluster membership and its network.
1102 For further information about it, check the corosync.conf man page:
1108 For node membership you should always use the `pvecm` tool provided by {pve}.
1109 You may have to edit the configuration file manually for other changes.
1110 Here are a few best practice tips for doing this.
1112 [[pvecm_edit_corosync_conf]]
1116 Editing the corosync.conf file is not always very straightforward. There are
1117 two on each cluster node, one in `/etc/pve/corosync.conf` and the other in
1118 `/etc/corosync/corosync.conf`. Editing the one in our cluster file system will
1119 propagate the changes to the local one, but not vice versa.
1121 The configuration will get updated automatically as soon as the file changes.
1122 This means changes which can be integrated in a running corosync will take
1123 effect immediately. So you should always make a copy and edit that instead, to
1124 avoid triggering some unwanted changes by an in-between safe.
1128 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
1131 Then open the config file with your favorite editor, `nano` and `vim.tiny` are
1132 preinstalled on any {pve} node for example.
1134 NOTE: Always increment the 'config_version' number on configuration changes,
1135 omitting this can lead to problems.
1137 After making the necessary changes create another copy of the current working
1138 configuration file. This serves as a backup if the new configuration fails to
1139 apply or makes problems in other ways.
1143 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak
1146 Then move the new configuration file over the old one:
1149 mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
1152 You may check with the commands
1155 systemctl status corosync
1156 journalctl -b -u corosync
1159 If the change could be applied automatically. If not you may have to restart the
1160 corosync service via:
1163 systemctl restart corosync
1166 On errors check the troubleshooting section below.
1171 Issue: 'quorum.expected_votes must be configured'
1172 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1174 When corosync starts to fail and you get the following message in the system log:
1178 corosync[1647]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
1179 corosync[1647]: [SERV ] Service engine 'corosync_quorum' failed to load for reason
1180 'configuration error: nodelist or quorum.expected_votes must be configured!'
1184 It means that the hostname you set for corosync 'ringX_addr' in the
1185 configuration could not be resolved.
1187 Write Configuration When Not Quorate
1188 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1190 If you need to change '/etc/pve/corosync.conf' on an node with no quorum, and you
1191 know what you do, use:
1197 This sets the expected vote count to 1 and makes the cluster quorate. You can
1198 now fix your configuration, or revert it back to the last working backup.
1200 This is not enough if corosync cannot start anymore. Here it is best to edit the
1201 local copy of the corosync configuration in '/etc/corosync/corosync.conf' so
1202 that corosync can start again. Ensure that on all nodes this configuration has
1203 the same content to avoid split brains. If you are not sure what went wrong
1204 it's best to ask the Proxmox Community to help you.
1207 [[pvecm_corosync_conf_glossary]]
1208 Corosync Configuration Glossary
1209 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1212 This names the different link addresses for the kronosnet connections between
1219 It is obvious that a cluster is not quorate when all nodes are
1220 offline. This is a common case after a power failure.
1222 NOTE: It is always a good idea to use an uninterruptible power supply
1223 (``UPS'', also called ``battery backup'') to avoid this state, especially if
1226 On node startup, the `pve-guests` service is started and waits for
1227 quorum. Once quorate, it starts all guests which have the `onboot`
1230 When you turn on nodes, or when power comes back after power failure,
1231 it is likely that some nodes boots faster than others. Please keep in
1232 mind that guest startup is delayed until you reach quorum.
1238 Migrating virtual guests to other nodes is a useful feature in a
1239 cluster. There are settings to control the behavior of such
1240 migrations. This can be done via the configuration file
1241 `datacenter.cfg` or for a specific migration via API or command line
1244 It makes a difference if a Guest is online or offline, or if it has
1245 local resources (like a local disk).
1247 For Details about Virtual Machine Migration see the
1248 xref:qm_migration[QEMU/KVM Migration Chapter].
1250 For Details about Container Migration see the
1251 xref:pct_migration[Container Migration Chapter].
1256 The migration type defines if the migration data should be sent over an
1257 encrypted (`secure`) channel or an unencrypted (`insecure`) one.
1258 Setting the migration type to insecure means that the RAM content of a
1259 virtual guest gets also transferred unencrypted, which can lead to
1260 information disclosure of critical data from inside the guest (for
1261 example passwords or encryption keys).
1263 Therefore, we strongly recommend using the secure channel if you do
1264 not have full control over the network and can not guarantee that no
1265 one is eavesdropping on it.
1267 NOTE: Storage migration does not follow this setting. Currently, it
1268 always sends the storage content over a secure channel.
1270 Encryption requires a lot of computing power, so this setting is often
1271 changed to "unsafe" to achieve better performance. The impact on
1272 modern systems is lower because they implement AES encryption in
1273 hardware. The performance impact is particularly evident in fast
1274 networks where you can transfer 10 Gbps or more.
1279 By default, {pve} uses the network in which cluster communication
1280 takes place to send the migration traffic. This is not optimal because
1281 sensitive cluster traffic can be disrupted and this network may not
1282 have the best bandwidth available on the node.
1284 Setting the migration network parameter allows the use of a dedicated
1285 network for the entire migration traffic. In addition to the memory,
1286 this also affects the storage traffic for offline migrations.
1288 The migration network is set as a network in the CIDR notation. This
1289 has the advantage that you do not have to set individual IP addresses
1290 for each node. {pve} can determine the real address on the
1291 destination node from the network specified in the CIDR form. To
1292 enable this, the network must be specified so that each node has one,
1293 but only one IP in the respective network.
1298 We assume that we have a three-node setup with three separate
1299 networks. One for public communication with the Internet, one for
1300 cluster communication and a very fast one, which we want to use as a
1301 dedicated network for migration.
1303 A network configuration for such a setup might look as follows:
1306 iface eno1 inet manual
1310 iface vmbr0 inet static
1312 netmask 255.255.250.0
1320 iface eno2 inet static
1322 netmask 255.255.255.0
1326 iface eno3 inet static
1328 netmask 255.255.255.0
1331 Here, we will use the network 10.1.2.0/24 as a migration network. For
1332 a single migration, you can do this using the `migration_network`
1333 parameter of the command line tool:
1336 # qm migrate 106 tre --online --migration_network 10.1.2.0/24
1339 To configure this as the default network for all migrations in the
1340 cluster, set the `migration` property of the `/etc/pve/datacenter.cfg`
1344 # use dedicated migration network
1345 migration: secure,network=10.1.2.0/24
1348 NOTE: The migration type must always be set when the migration network
1349 gets set in `/etc/pve/datacenter.cfg`.
1353 include::pve-copyright.adoc[]