Difference between revisions of "Cluster Manager"

From Proxmox VE
Jump to: navigation, search
 
Line 24: Line 24:
 
Cluster-wide services like firewall and HA
 
Cluster-wide services like firewall and HA
 
Requirements
 
Requirements
All nodes must be in the same network as corosync uses IP Multicast
+
All nodes must be able to connect to each other via UDP ports 5404 and 5405
to communicate between nodes (also see
+
for corosync to work.
Corosync Cluster Engine). Corosync uses UDP
 
ports 5404 and 5405 for cluster communication.
 
Some switches do not support IP multicast by default and must be
 
manually enabled first.
 
 
Date and time have to be synchronized.
 
Date and time have to be synchronized.
 
SSH tunnel on TCP port 22 between nodes is used.
 
SSH tunnel on TCP port 22 between nodes is used.
Line 43: Line 39:
 
production configuration and should only used temporarily during upgrading the
 
production configuration and should only used temporarily during upgrading the
 
whole cluster from one to another major version.
 
whole cluster from one to another major version.
 +
Running a cluster of Proxmox VE 6.x with earlier versions is not possible. The
 +
cluster protocol (corosync) between Proxmox VE 6.x and earlier versions changed
 +
fundamentally. The corosync 3 packages for Proxmox VE 5.4 are only intended for the
 +
upgrade procedure to Proxmox VE 6.0.
 
Preparing Nodes
 
Preparing Nodes
 
First, install Proxmox VE on all nodes. Make sure that each node is
 
First, install Proxmox VE on all nodes. Make sure that each node is
Line 50: Line 50:
 
ssh) or the API, which we have a GUI implementation for (Datacenter →
 
ssh) or the API, which we have a GUI implementation for (Datacenter →
 
Cluster).
 
Cluster).
While it’s often common use to reference all other nodenames in /etc/hosts
+
While it’s common to reference all nodenames and their IPs in /etc/hosts (or
with their IP this is not strictly necessary for a cluster, which normally uses
+
make their names resolvable through other means), this is not necessary for a
multicast, to work. It maybe useful as you then can connect from one node to
+
cluster to work. It may be useful however, as you can then connect from one node
the other with SSH through the easier to remember node name.
+
to the other with SSH via the easier to remember node name (see also
 +
Link Address Types). Note that we always
 +
recommend to reference nodes by their IP addresses in the cluster configuration.
 
Create the Cluster
 
Create the Cluster
 
Login via ssh to the first Proxmox VE node. Use a unique name for your cluster.
 
Login via ssh to the first Proxmox VE node. Use a unique name for your cluster.
Line 59: Line 61:
 
node names.
 
node names.
 
  hp1# pvecm create CLUSTERNAME
 
  hp1# pvecm create CLUSTERNAME
The cluster name is used to compute the default multicast address.
+
It is possible to create multiple clusters in the same physical or logical
Please use unique cluster names if you run more than one cluster inside your
+
network. Use unique cluster names if you do so. To avoid human confusion, it is
network. To avoid human confusion, it is also recommended to choose different
+
also recommended to choose different names even if clusters do not share the
names even if clusters do not share the cluster network.
+
cluster network.
 
To check the state of your cluster use:
 
To check the state of your cluster use:
 
  hp1# pvecm status
 
  hp1# pvecm status
 
Multiple Clusters In Same Network
 
Multiple Clusters In Same Network
 
It is possible to create multiple clusters in the same physical or logical
 
It is possible to create multiple clusters in the same physical or logical
network. Each cluster must have a unique name, which is used to generate the
+
network. Each such cluster must have a unique name, this does not only helps
cluster’s multicast group address. As long as no duplicate cluster names are
+
admins to distinguish on which cluster they currently operate, it is also
configured in one network segment, the different clusters won’t interfere with
+
required to avoid possible clashes in the cluster communication stack.
each other.
+
While the bandwidth requirement of a corosync cluster is relatively low, the
If multiple clusters operate in a single network it may be beneficial to setup
+
latency of packages and the package per second (PPS) rate is the limiting
an IGMP querier and enable IGMP Snooping in said network. This may reduce the
+
factor. Different clusters in the same network can compete with each other for
load of the network significantly because multicast packets are only delivered
+
these resources, so it may still make sense to use separate physical network
to endpoints of the respective member nodes.
+
infrastructure for bigger clusters.
 
Adding Nodes to the Cluster
 
Adding Nodes to the Cluster
 
Login via ssh to the node you want to add.
 
Login via ssh to the node you want to add.
 
  hp2# pvecm add IP-ADDRESS-CLUSTER
 
  hp2# pvecm add IP-ADDRESS-CLUSTER
For IP-ADDRESS-CLUSTER use the IP from an existing cluster node.
+
For IP-ADDRESS-CLUSTER use the IP or hostname of an existing cluster node.
 +
An IP address is recommended (see Link Address Types).
 
A new node cannot hold any VMs, because you would get
 
A new node cannot hold any VMs, because you would get
 
conflicts about identical VM IDs. Also, all existing configuration in
 
conflicts about identical VM IDs. Also, all existing configuration in
Line 84: Line 87:
 
workaround, use vzdump to backup and restore to a different VMID after
 
workaround, use vzdump to backup and restore to a different VMID after
 
adding the node to the cluster.
 
adding the node to the cluster.
To check the state of cluster:
+
To check the state of the cluster use:
 
  # pvecm status
 
  # pvecm status
 
Cluster status after adding 4 nodes
 
Cluster status after adding 4 nodes
Line 94: Line 97:
 
Nodes:            4
 
Nodes:            4
 
Node ID:          0x00000001
 
Node ID:          0x00000001
Ring ID:          1928
+
Ring ID:          1/8
 
Quorate:          Yes
 
Quorate:          Yes
 
Votequorum information
 
Votequorum information
Line 123: Line 126:
 
Adding Nodes With Separated Cluster Network
 
Adding Nodes With Separated Cluster Network
 
When adding a node to a cluster with a separated cluster network you need to
 
When adding a node to a cluster with a separated cluster network you need to
use the ringX_addr parameters to set the nodes address on those networks:
+
use the link0 parameter to set the nodes address on that network:
pvecm add IP-ADDRESS-CLUSTER -ring0_addr IP-ADDRESS-RING0
+
pvecm add IP-ADDRESS-CLUSTER -link0 LOCAL-IP-ADDRESS-LINK0
If you want to use the Redundant Ring Protocol you will also want to pass the
+
If you want to use the built-in redundancy of the
ring1_addr parameter.
+
kronosnet transport layer, also use the link1 parameter.
 
Remove a Cluster Node
 
Remove a Cluster Node
 
Read carefully the procedure before proceeding, as it could
 
Read carefully the procedure before proceeding, as it could
Line 163: Line 166:
 
Nodes:            3
 
Nodes:            3
 
Node ID:          0x00000001
 
Node ID:          0x00000001
Ring ID:          1992
+
Ring ID:          1/8
 
Quorate:          Yes
 
Quorate:          Yes
 
Votequorum information
 
Votequorum information
Line 178: Line 181:
 
0x00000002          1 192.168.15.91
 
0x00000002          1 192.168.15.91
 
0x00000003          1 192.168.15.92
 
0x00000003          1 192.168.15.92
If, for whatever reason, you want that this server joins the same
+
If, for whatever reason, you want this server to join the same cluster again,
cluster again, you have to
+
you have to
 
reinstall Proxmox VE on it from scratch
 
reinstall Proxmox VE on it from scratch
 
then join it, as explained in the previous section.
 
then join it, as explained in the previous section.
 +
After removal of the node, its SSH fingerprint will still reside in the
 +
known_hosts of the other nodes. If you receive an SSH error after rejoining
 +
a node with the same IP or hostname, run pvecm updatecerts once on the
 +
re-added node to update its fingerprint cluster wide.
 
Separate A Node Without Reinstalling
 
Separate A Node Without Reinstalling
 
This is not the recommended method, proceed with caution. Use the
 
This is not the recommended method, proceed with caution. Use the
Line 192: Line 199:
 
boundary. Further, it may also lead to VMID conflicts.
 
boundary. Further, it may also lead to VMID conflicts.
 
Its suggested that you create a new storage where only the node which you want
 
Its suggested that you create a new storage where only the node which you want
to separate has access. This can be an new export on your NFS or a new Ceph
+
to separate has access. This can be a new export on your NFS or a new Ceph
 
pool, to name a few examples. Its just important that the exact same storage
 
pool, to name a few examples. Its just important that the exact same storage
 
does not gets accessed by multiple clusters. After setting this storage up move
 
does not gets accessed by multiple clusters. After setting this storage up move
 
all data from the node and its VMs to it. Then you are ready to separate the
 
all data from the node and its VMs to it. Then you are ready to separate the
 
node from the cluster.
 
node from the cluster.
Ensure all shared resources are cleanly separated! You will run into
+
Ensure all shared resources are cleanly separated! Otherwise you will
conflicts and problems else.
+
run into conflicts and problems.
 
First stop the corosync and the pve-cluster services on the node:
 
First stop the corosync and the pve-cluster services on the node:
 
systemctl stop pve-cluster
 
systemctl stop pve-cluster
Line 243: Line 250:
 
Cluster Network
 
Cluster Network
 
The cluster network is the core of a cluster. All messages sent over it have to
 
The cluster network is the core of a cluster. All messages sent over it have to
be delivered reliable to all nodes in their respective order. In Proxmox VE this
+
be delivered reliably to all nodes in their respective order. In Proxmox VE this
part is done by corosync, an implementation of a high performance low overhead
+
part is done by corosync, an implementation of a high performance, low overhead
 
high availability development toolkit. It serves our decentralized
 
high availability development toolkit. It serves our decentralized
 
configuration file system (pmxcfs).
 
configuration file system (pmxcfs).
 
Network Requirements
 
Network Requirements
 
This needs a reliable network with latencies under 2 milliseconds (LAN
 
This needs a reliable network with latencies under 2 milliseconds (LAN
performance) to work properly. While corosync can also use unicast for
+
performance) to work properly. The network should not be used heavily by other
communication between nodes its highly recommended to have a multicast
+
members, ideally corosync runs on its own network. Do not use a shared network
capable network. The network should not be used heavily by other members,
+
for corosync and storage (except as a potential low-priority fallback in a
ideally corosync runs on its own network.
+
redundant configuration).
never share it with network where storage communicates too.
+
Before setting up a cluster, it is good practice to check if the network is fit
Before setting up a cluster it is good practice to check if the network is fit
+
for that purpose. To make sure the nodes can connect to each other on the
for that purpose.
+
cluster network, you can test the connectivity between them with the ping
Ensure that all nodes are in the same subnet. This must only be true for the
+
tool.
  network interfaces used for cluster communication (corosync).
+
If the Proxmox VE firewall is enabled, ACCEPT rules for corosync will automatically
Ensure all nodes can reach each other over those interfaces, using ping is
+
be generated - no manual action is required.
  enough for a basic test.
+
Corosync used Multicast before version 3.0 (introduced in Proxmox VE 6.0).
Ensure that multicast works in general and a high package rates. This can be
+
Modern versions rely on Kronosnet for cluster
  done with the omping tool. The final "%loss" number should be < 1%.
+
communication, which, for now, only supports regular UDP unicast.
omping -c 10000 -i 0.001 -F -q NODE1-IP NODE2-IP ...
+
You can still enable Multicast or legacy unicast by setting your
Ensure that multicast communication works over an extended period of time.
+
transport to udp or udpu in your corosync.conf,
  This uncovers problems where IGMP snooping is activated on the network but
+
but keep in mind that this will disable all cryptography and redundancy support.
  no multicast querier is active. This test has a duration of around 10
+
This is therefore not recommended.
  minutes.
 
omping -c 600 -i 1 -q NODE1-IP NODE2-IP ...
 
Your network is not ready for clustering if any of these test fails. Recheck
 
your network configuration. Especially switches are notorious for having
 
multicast disabled by default or IGMP snooping enabled with no IGMP querier
 
active.
 
In smaller cluster its also an option to use unicast if you really cannot get
 
multicast to work.
 
 
Separate Cluster Network
 
Separate Cluster Network
When creating a cluster without any parameters the cluster network is generally
+
When creating a cluster without any parameters the corosync cluster network is
shared with the Web UI and the VMs and its traffic. Depending on your setup
+
generally shared with the Web UI and the VMs and their traffic. Depending on
even storage traffic may get sent over the same network. Its recommended to
+
your setup, even storage traffic may get sent over the same network. Its
change that, as corosync is a time critical real time application.
+
recommended to change that, as corosync is a time critical real time
 +
application.
 
Setting Up A New Network
 
Setting Up A New Network
First you have to setup a new network interface. It should be on a physical
+
First you have to set up a new network interface. It should be on a physically
 
separate network. Ensure that your network fulfills the
 
separate network. Ensure that your network fulfills the
 
cluster network requirements.
 
cluster network requirements.
 
Separate On Cluster Creation
 
Separate On Cluster Creation
This is possible through the ring0_addr and bindnet0_addr parameter of
+
This is possible via the linkX parameters of the pvecm create
the pvecm create command used for creating a new cluster.
+
command used for creating a new cluster.
If you have setup an additional NIC with a static address on 10.10.10.1/25
+
If you have set up an additional NIC with a static address on 10.10.10.1/25,
and want to send and receive all cluster communication over this interface
+
and want to send and receive all cluster communication over this interface,
 
you would execute:
 
you would execute:
pvecm create test --ring0_addr 10.10.10.1 --bindnet0_addr 10.10.10.0
+
pvecm create test --link0 10.10.10.1
 
To check if everything is working properly execute:
 
To check if everything is working properly execute:
 
systemctl status corosync
 
systemctl status corosync
Afterwards, proceed as descripted in the section to
+
Afterwards, proceed as described above to
 
add nodes with a separated cluster network.
 
add nodes with a separated cluster network.
 
Separate After Cluster Creation
 
Separate After Cluster Creation
You can do this also if you have already created a cluster and want to switch
+
You can do this if you have already created a cluster and want to switch
 
its communication to another network, without rebuilding the whole cluster.
 
its communication to another network, without rebuilding the whole cluster.
 
This change may lead to short durations of quorum loss in the cluster, as nodes
 
This change may lead to short durations of quorum loss in the cluster, as nodes
 
have to restart corosync and come up one after the other on the new network.
 
have to restart corosync and come up one after the other on the new network.
 
Check how to edit the corosync.conf file first.
 
Check how to edit the corosync.conf file first.
The open it and you should see a file similar to:
+
Then, open it and you should see a file similar to:
 
logging {
 
logging {
 
   debug: off
 
   debug: off
Line 329: Line 329:
 
}
 
}
 
totem {
 
totem {
   cluster_name: thomas-testcluster
+
   cluster_name: testcluster
 
   config_version: 3
 
   config_version: 3
   ip_version: ipv4
+
   ip_version: ipv4-6
 
   secauth: on
 
   secauth: on
 
   version: 2
 
   version: 2
 
   interface {
 
   interface {
     bindnetaddr: 192.168.30.50
+
     linknumber: 0
    ringnumber: 0
 
 
   }
 
   }
 
}
 
}
The first you want to do is add the name properties in the node entries if
+
ringX_addr actually specifies a corosync link address, the name "ring"
you do not see them already. Those must match the node name.
+
is a remnant of older corosync versions that is kept for backwards
Then replace the address from the ring0_addr properties with the new
+
compatibility.
addresses. You may use plain IP addresses or also hostnames here. If you use
+
The first thing you want to do is add the name properties in the node entries
hostnames ensure that they are resolvable from all nodes.
+
if you do not see them already. Those must match the node name.
In my example I want to switch my cluster communication to the 10.10.10.1/25
+
Then replace all addresses from the ring0_addr properties of all nodes with
network. So I replace all ring0_addr respectively. I also set the bindnetaddr
+
the new addresses. You may use plain IP addresses or hostnames here. If you use
in the totem section of the config to an address of the new network. It can be
+
hostnames ensure that they are resolvable from all nodes. (see also
any address from the subnet configured on the new network interface.
+
Link Address Types)
After you increased the config_version property the new configuration file
+
In this example, we want to switch the cluster communication to the
 +
10.10.10.1/25 network. So we replace all ring0_addr respectively.
 +
The exact same procedure can be used to change other ringX_addr values
 +
as well, although we recommend to not change multiple addresses at once, to make
 +
it easier to recover if something goes wrong.
 +
After we increase the config_version property, the new configuration file
 
should look like:
 
should look like:
 
logging {
 
logging {
Line 378: Line 382:
 
}
 
}
 
totem {
 
totem {
   cluster_name: thomas-testcluster
+
   cluster_name: testcluster
 
   config_version: 4
 
   config_version: 4
   ip_version: ipv4
+
   ip_version: ipv4-6
 
   secauth: on
 
   secauth: on
 
   version: 2
 
   version: 2
 
   interface {
 
   interface {
     bindnetaddr: 10.10.10.1
+
     linknumber: 0
    ringnumber: 0
 
 
   }
 
   }
 
}
 
}
Now after a final check whether all changed information is correct we save it
+
Then, after a final check if all changed information is correct, we save it and
and see again the edit corosync.conf file section to
+
once again follow the edit corosync.conf file
learn how to bring it in effect.
+
section to bring it into effect.
As our change cannot be enforced live from corosync we have to do an restart.
+
The changes will be applied live, so restarting corosync is not strictly
 +
necessary. If you changed other settings as well, or notice corosync
 +
complaining, you can optionally trigger a restart.
 
On a single node execute:
 
On a single node execute:
 
systemctl restart corosync
 
systemctl restart corosync
Line 398: Line 403:
 
If corosync runs again correct restart corosync also on all other nodes.
 
If corosync runs again correct restart corosync also on all other nodes.
 
They will then join the cluster membership one by one on the new network.
 
They will then join the cluster membership one by one on the new network.
Redundant Ring Protocol
+
Corosync addresses
To avoid a single point of failure you should implement counter measurements.
+
A corosync link address (for backwards compatibility denoted by ringX_addr in
This can be on the hardware and operating system level through network bonding.
+
corosync.conf) can be specified in two ways:
Corosync itself offers also a possibility to add redundancy through the so
+
IPv4/v6 addresses will be used directly. They are recommended, since they
called Redundant Ring Protocol. This protocol allows running a second totem
+
are static and usually not changed carelessly.
ring on another network, this network should be physically separated from the
+
Hostnames will be resolved using getaddrinfo, which means that per
other rings network to actually increase availability.
+
default, IPv6 addresses will be used first, if available (see also
RRP On Cluster Creation
+
man gai.conf). Keep this in mind, especially when upgrading an existing
The pvecm create command provides the additional parameters bindnetX_addr,
+
cluster to IPv6.
ringX_addr and rrp_mode, can be used for RRP configuration.
+
Hostnames should be used with care, since the address they
See the glossary if you do not know what each parameter means.
+
resolve to can be changed without touching corosync or the node it runs on -
So if you have two networks, one on the 10.10.10.1/24 and the other on the
+
which may lead to a situation where an address is changed without thinking
10.10.20.1/24 subnet you would execute:
+
about implications for corosync.
pvecm create CLUSTERNAME -bindnet0_addr 10.10.10.1 -ring0_addr 10.10.10.1 \
+
A seperate, static hostname specifically for corosync is recommended, if
-bindnet1_addr 10.10.20.1 -ring1_addr 10.10.20.1
+
hostnames are preferred. Also, make sure that every node in the cluster can
RRP On Existing Clusters
+
resolve all hostnames correctly.
You will take similar steps as described in
+
Since Proxmox VE 5.1, while supported, hostnames will be resolved at the time of
separating the cluster network to
+
entry. Only the resolved IP is then saved to the configuration.
enable RRP on an already running cluster. The single difference is, that you
+
Nodes that joined the cluster on earlier versions likely still use their
will add ring1 and use it instead of ring0.
+
unresolved hostname in corosync.conf. It might be a good idea to replace
First add a new interface subsection in the totem section, set its
+
them with IPs or a seperate hostname, as mentioned above.
ringnumber property to 1. Set the interfaces bindnetaddr property to an
+
Corosync Redundancy
address of the subnet you have configured for your new ring.
+
Corosync supports redundant networking via its integrated kronosnet layer by
Further set the rrp_mode to passive, this is the only stable mode.
+
default (it is not supported on the legacy udp/udpu transports). It can be
Then add to each node entry in the nodelist section its new ring1_addr
+
enabled by specifying more than one link address, either via the --linkX
property with the nodes additional ring address.
+
parameters of pvecm (while creating a cluster or adding a new node) or by
So if you have two networks, one on the 10.10.10.1/24 and the other on the
+
specifying more than one ringX_addr in corosync.conf.
10.10.20.1/24 subnet, the final configuration file should look like:
+
To provide useful failover, every link should be on its own
totem {
+
physical network connection.
   cluster_name: tweak
+
Links are used according to a priority setting. You can configure this priority
   config_version: 9
+
by setting knet_link_priority in the corresponding interface section in
  ip_version: ipv4
+
corosync.conf, or, preferrably, using the priority parameter when creating
   rrp_mode: passive
+
your cluster with pvecm:
  secauth: on
+
# pvecm create CLUSTERNAME --link0 10.10.10.1,priority=20 --link1 10.20.20.1,priority=15
  version: 2
+
This would cause link1 to be used first, since it has the lower priority.
  interface {
+
If no priorities are configured manually (or two links have the same priority),
     bindnetaddr: 10.10.10.1
+
links will be used in order of their number, with the lower number having higher
     ringnumber: 0
+
priority.
 +
Even if all links are working, only the one with the highest priority will see
 +
corosync traffic. Link priorities cannot be mixed, i.e. links with different
 +
priorities will not be able to communicate with each other.
 +
Since lower priority links will not see traffic unless all higher priorities
 +
have failed, it becomes a useful strategy to specify even networks used for
 +
other tasks (VMs, storage, etc…) as low-priority links. If worst comes to
 +
worst, a higher-latency or more congested connection might be better than no
 +
connection at all.
 +
Adding Redundant Links To An Existing Cluster
 +
To add a new link to a running configuration, first check how to
 +
edit the corosync.conf file.
 +
Then, add a new ringX_addr to every node in the nodelist section. Make
 +
sure that your X is the same for every node you add it to, and that it is
 +
unique for each node.
 +
Lastly, add a new interface, as shown below, to your totem
 +
section, replacing X with your link number chosen above.
 +
Assuming you added a link with number 1, the new configuration file could look
 +
like this:
 +
logging {
 +
   debug: off
 +
   to_syslog: yes
 +
}
 +
nodelist {
 +
   node {
 +
    name: due
 +
    nodeid: 2
 +
    quorum_votes: 1
 +
     ring0_addr: 10.10.10.2
 +
     ring1_addr: 10.20.20.2
 
   }
 
   }
   interface {
+
   node {
     bindnetaddr: 10.10.20.1
+
     name: tre
     ringnumber: 1
+
    nodeid: 3
 +
    quorum_votes: 1
 +
    ring0_addr: 10.10.10.3
 +
     ring1_addr: 10.20.20.3
 
   }
 
   }
}
 
nodelist {
 
 
   node {
 
   node {
     name: pvecm1
+
     name: uno
 
     nodeid: 1
 
     nodeid: 1
 
     quorum_votes: 1
 
     quorum_votes: 1
 
     ring0_addr: 10.10.10.1
 
     ring0_addr: 10.10.10.1
     ring1_addr: 10.10.20.1
+
     ring1_addr: 10.20.20.1
 +
  }
 +
}
 +
quorum {
 +
  provider: corosync_votequorum
 +
}
 +
totem {
 +
  cluster_name: testcluster
 +
  config_version: 4
 +
  ip_version: ipv4-6
 +
  secauth: on
 +
  version: 2
 +
  interface {
 +
    linknumber: 0
 
   }
 
   }
node {
+
  interface {
     name: pvecm2
+
     linknumber: 1
    nodeid: 2
 
    quorum_votes: 1
 
    ring0_addr: 10.10.10.2
 
    ring1_addr: 10.10.20.2
 
 
   }
 
   }
  [...] # other cluster nodes here
 
 
}
 
}
[...] # other remaining config sections here
+
The new link will be enabled as soon as you follow the last steps to
Bring it in effect like described in the
+
edit the corosync.conf file. A restart should not
edit the corosync.conf file section.
+
be necessary. You can check that corosync loaded the new link using:
This is a change which cannot take live in effect and needs at least a restart
+
journalctl -b -u corosync
of corosync. Recommended is a restart of the whole cluster.
+
It might be a good idea to test the new link by temporarily disconnecting the
If you cannot reboot the whole cluster ensure no High Availability services are
+
old link on one node and making sure that its status remains online while
configured and the stop the corosync service on all nodes. After corosync is
+
disconnected:
stopped on all nodes start it one after the other again.
+
pvecm status
 +
If you see a healthy cluster state, it means that your new link is being used.
 
Corosync External Vote Support
 
Corosync External Vote Support
 
This section describes a way to deploy an external voter in a Proxmox VE cluster.
 
This section describes a way to deploy an external voter in a Proxmox VE cluster.
Line 497: Line 541:
 
available through their respective package manager.
 
available through their respective package manager.
 
In contrast to corosync itself, a QDevice connects to the cluster over
 
In contrast to corosync itself, a QDevice connects to the cluster over
TCP/IP and thus does not need a multicast capable network between itself and
+
TCP/IP. The daemon may even run outside of the clusters LAN and can have longer
the cluster. In fact the daemon may run outside of the LAN and can have
+
latencies than 2 ms.
longer latencies than 2 ms.
 
 
Supported Setups
 
Supported Setups
 
We support QDevices for clusters with an even number of nodes and recommend
 
We support QDevices for clusters with an even number of nodes and recommend
Line 528: Line 571:
 
QDevice-Net Setup
 
QDevice-Net Setup
 
We recommend to run any daemon which provides votes to corosync-qdevice as an
 
We recommend to run any daemon which provides votes to corosync-qdevice as an
unprivileged user. Proxmox VE and Debian Stretch provide a package which is
+
unprivileged user. Proxmox VE and Debian provides a package which is already
already configured to do so.
+
configured to do so.
 
The traffic between the daemon and the cluster must be encrypted to ensure a
 
The traffic between the daemon and the cluster must be encrypted to ensure a
 
safe and secure QDevice integration in Proxmox VE.
 
safe and secure QDevice integration in Proxmox VE.
Line 577: Line 620:
 
pve# pvecm qdevice remove
 
pve# pvecm qdevice remove
 
Corosync Configuration
 
Corosync Configuration
The /etc/pve/corosync.conf file plays a central role in Proxmox VE cluster. It
+
The /etc/pve/corosync.conf file plays a central role in a Proxmox VE cluster. It
controls the cluster member ship and its network.
+
controls the cluster membership and its network.
For reading more about it check the corosync.conf man page:
+
For further information about it, check the corosync.conf man page:
 
man corosync.conf
 
man corosync.conf
 
For node membership you should always use the pvecm tool provided by Proxmox VE.
 
For node membership you should always use the pvecm tool provided by Proxmox VE.
Line 585: Line 628:
 
Here are a few best practice tips for doing this.
 
Here are a few best practice tips for doing this.
 
Edit corosync.conf
 
Edit corosync.conf
Editing the corosync.conf file can be not always straight forward. There are
+
Editing the corosync.conf file is not always very straightforward. There are
two on each cluster, one in /etc/pve/corosync.conf and the other in
+
two on each cluster node, one in /etc/pve/corosync.conf and the other in
 
/etc/corosync/corosync.conf. Editing the one in our cluster file system will
 
/etc/corosync/corosync.conf. Editing the one in our cluster file system will
 
propagate the changes to the local one, but not vice versa.
 
propagate the changes to the local one, but not vice versa.
 
The configuration will get updated automatically as soon as the file changes.
 
The configuration will get updated automatically as soon as the file changes.
 
This means changes which can be integrated in a running corosync will take
 
This means changes which can be integrated in a running corosync will take
instantly effect. So you should always make a copy and edit that instead, to
+
effect immediately. So you should always make a copy and edit that instead, to
avoid triggering some unwanted changes by an in between safe.
+
avoid triggering some unwanted changes by an in-between safe.
 
cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
 
cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
Then open the Config file with your favorite editor, nano and vim.tiny are
+
Then open the config file with your favorite editor, nano and vim.tiny are
preinstalled on Proxmox VE for example.
+
preinstalled on any Proxmox VE node for example.
 
Always increment the config_version number on configuration changes,
 
Always increment the config_version number on configuration changes,
 
omitting this can lead to problems.
 
omitting this can lead to problems.
Line 607: Line 650:
 
systemctl status corosync
 
systemctl status corosync
 
journalctl -b -u corosync
 
journalctl -b -u corosync
If the change could applied automatically. If not you may have to restart the
+
If the change could be applied automatically. If not you may have to restart the
 
corosync service via:
 
corosync service via:
 
systemctl restart corosync
 
systemctl restart corosync
Line 634: Line 677:
 
Corosync Configuration Glossary
 
Corosync Configuration Glossary
 
ringX_addr
 
ringX_addr
This names the different ring addresses for the corosync totem rings used for
+
This names the different link addresses for the kronosnet connections between
the cluster communication.
+
nodes.
bindnetaddr
 
Defines to which interface the ring should bind to. It may be any address of
 
the subnet configured on the interface we want to use. In general its the
 
recommended to just use an address a node uses on this interface.
 
rrp_mode
 
Specifies the mode of the redundant ring protocol and may be passive, active or
 
none. Note that use of active is highly experimental and not official
 
supported. Passive is the preferred mode, it may double the cluster
 
communication throughput and increases availability.
 
 
Cluster Cold Start
 
Cluster Cold Start
 
It is obvious that a cluster is not quorate when all nodes are
 
It is obvious that a cluster is not quorate when all nodes are
Line 666: Line 700:
 
local resources (like a local disk).
 
local resources (like a local disk).
 
For Details about Virtual Machine Migration see the
 
For Details about Virtual Machine Migration see the
QEMU/KVM Migration Chapter
+
QEMU/KVM Migration Chapter.
 
For Details about Container Migration see the
 
For Details about Container Migration see the
Container Migration Chapter
+
Container Migration Chapter.
 
Migration Type
 
Migration Type
 
The migration type defines if the migration data should be sent over an
 
The migration type defines if the migration data should be sent over an

Latest revision as of 11:23, 16 July 2019

The Proxmox VE cluster manager pvecm is a tool to create a group of physical servers. Such a group is called a cluster. We use the Corosync Cluster Engine for reliable group communication, and such clusters can consist of up to 32 physical nodes (probably more, dependent on network latency).

pvecm can be used to create a new cluster, join nodes to a cluster, leave the cluster, get status information and do various other cluster related tasks. The Proxmox Cluster File System (“pmxcfs”) is used to transparently distribute the cluster configuration to all cluster nodes.

Grouping nodes into a cluster has the following advantages:

  • Centralized, web based management

  • Multi-master clusters: each node can do all management task

  • pmxcfs: database-driven file system for storing configuration files, replicated in real-time on all nodes using corosync.

  • Easy migration of virtual machines and containers between physical hosts

  • Fast deployment

  • Cluster-wide services like firewall and HA

Requirements

  • All nodes must be able to connect to each other via UDP ports 5404 and 5405 for corosync to work.

  • Date and time have to be synchronized.

  • SSH tunnel on TCP port 22 between nodes is used.

  • If you are interested in High Availability, you need to have at least three nodes for reliable quorum. All nodes should have the same version.

  • We recommend a dedicated NIC for the cluster traffic, especially if you use shared storage.

  • Root password of a cluster node is required for adding nodes.

Note It is not possible to mix Proxmox VE 3.x and earlier with Proxmox VE 4.X cluster nodes.
Note While it’s possible for Proxmox VE 4.4 and Proxmox VE 5.0 this is not supported as production configuration and should only used temporarily during upgrading the whole cluster from one to another major version.
Note Running a cluster of Proxmox VE 6.x with earlier versions is not possible. The cluster protocol (corosync) between Proxmox VE 6.x and earlier versions changed fundamentally. The corosync 3 packages for Proxmox VE 5.4 are only intended for the upgrade procedure to Proxmox VE 6.0.

Preparing Nodes

First, install Proxmox VE on all nodes. Make sure that each node is installed with the final hostname and IP configuration. Changing the hostname and IP is not possible after cluster creation.

Currently the cluster creation can either be done on the console (login via ssh) or the API, which we have a GUI implementation for (Datacenter → Cluster).

While it’s common to reference all nodenames and their IPs in /etc/hosts (or make their names resolvable through other means), this is not necessary for a cluster to work. It may be useful however, as you can then connect from one node to the other with SSH via the easier to remember node name (see also Link Address Types). Note that we always recommend to reference nodes by their IP addresses in the cluster configuration.

Create the Cluster

Login via ssh to the first Proxmox VE node. Use a unique name for your cluster. This name cannot be changed later. The cluster name follows the same rules as node names.

 hp1# pvecm create CLUSTERNAME
Note It is possible to create multiple clusters in the same physical or logical network. Use unique cluster names if you do so. To avoid human confusion, it is also recommended to choose different names even if clusters do not share the cluster network.

To check the state of your cluster use:

 hp1# pvecm status

Multiple Clusters In Same Network

It is possible to create multiple clusters in the same physical or logical network. Each such cluster must have a unique name, this does not only helps admins to distinguish on which cluster they currently operate, it is also required to avoid possible clashes in the cluster communication stack.

While the bandwidth requirement of a corosync cluster is relatively low, the latency of packages and the package per second (PPS) rate is the limiting factor. Different clusters in the same network can compete with each other for these resources, so it may still make sense to use separate physical network infrastructure for bigger clusters.

Adding Nodes to the Cluster

Login via ssh to the node you want to add.

 hp2# pvecm add IP-ADDRESS-CLUSTER

For IP-ADDRESS-CLUSTER use the IP or hostname of an existing cluster node. An IP address is recommended (see Link Address Types).

Caution A new node cannot hold any VMs, because you would get conflicts about identical VM IDs. Also, all existing configuration in /etc/pve is overwritten when you join a new node to the cluster. To workaround, use vzdump to backup and restore to a different VMID after adding the node to the cluster.

To check the state of the cluster use:

 # pvecm status
Cluster status after adding 4 nodes
hp2# pvecm status
Quorum information
~~~~~~~~~~~~~~~~~~
Date:             Mon Apr 20 12:30:13 2015
Quorum provider:  corosync_votequorum
Nodes:            4
Node ID:          0x00000001
Ring ID:          1/8
Quorate:          Yes

Votequorum information
~~~~~~~~~~~~~~~~~~~~~~
Expected votes:   4
Highest expected: 4
Total votes:      4
Quorum:           3
Flags:            Quorate

Membership information
~~~~~~~~~~~~~~~~~~~~~~
    Nodeid      Votes Name
0x00000001          1 192.168.15.91
0x00000002          1 192.168.15.92 (local)
0x00000003          1 192.168.15.93
0x00000004          1 192.168.15.94

If you only want the list of all nodes use:

 # pvecm nodes
List nodes in a cluster
hp2# pvecm nodes

Membership information
~~~~~~~~~~~~~~~~~~~~~~
    Nodeid      Votes Name
         1          1 hp1
         2          1 hp2 (local)
         3          1 hp3
         4          1 hp4

Adding Nodes With Separated Cluster Network

When adding a node to a cluster with a separated cluster network you need to use the link0 parameter to set the nodes address on that network:

pvecm add IP-ADDRESS-CLUSTER -link0 LOCAL-IP-ADDRESS-LINK0

If you want to use the built-in redundancy of the kronosnet transport layer, also use the link1 parameter.

Remove a Cluster Node

Caution Read carefully the procedure before proceeding, as it could not be what you want or need.

Move all virtual machines from the node. Make sure you have no local data or backups you want to keep, or save them accordingly. In the following example we will remove the node hp4 from the cluster.

Log in to a different cluster node (not hp4), and issue a pvecm nodes command to identify the node ID to remove:

hp1# pvecm nodes

Membership information
~~~~~~~~~~~~~~~~~~~~~~
    Nodeid      Votes Name
         1          1 hp1 (local)
         2          1 hp2
         3          1 hp3
         4          1 hp4

At this point you must power off hp4 and make sure that it will not power on again (in the network) as it is.

Important As said above, it is critical to power off the node before removal, and make sure that it will never power on again (in the existing cluster network) as it is. If you power on the node as it is, your cluster will be screwed up and it could be difficult to restore a clean cluster state.

After powering off the node hp4, we can safely remove it from the cluster.

 hp1# pvecm delnode hp4

If the operation succeeds no output is returned, just check the node list again with pvecm nodes or pvecm status. You should see something like:

hp1# pvecm status

Quorum information
~~~~~~~~~~~~~~~~~~
Date:             Mon Apr 20 12:44:28 2015
Quorum provider:  corosync_votequorum
Nodes:            3
Node ID:          0x00000001
Ring ID:          1/8
Quorate:          Yes

Votequorum information
~~~~~~~~~~~~~~~~~~~~~~
Expected votes:   3
Highest expected: 3
Total votes:      3
Quorum:           2
Flags:            Quorate

Membership information
~~~~~~~~~~~~~~~~~~~~~~
    Nodeid      Votes Name
0x00000001          1 192.168.15.90 (local)
0x00000002          1 192.168.15.91
0x00000003          1 192.168.15.92

If, for whatever reason, you want this server to join the same cluster again, you have to

  • reinstall Proxmox VE on it from scratch

  • then join it, as explained in the previous section.

Note After removal of the node, its SSH fingerprint will still reside in the known_hosts of the other nodes. If you receive an SSH error after rejoining a node with the same IP or hostname, run pvecm updatecerts once on the re-added node to update its fingerprint cluster wide.

Separate A Node Without Reinstalling

Caution This is not the recommended method, proceed with caution. Use the above mentioned method if you’re unsure.

You can also separate a node from a cluster without reinstalling it from scratch. But after removing the node from the cluster it will still have access to the shared storages! This must be resolved before you start removing the node from the cluster. A Proxmox VE cluster cannot share the exact same storage with another cluster, as storage locking doesn’t work over cluster boundary. Further, it may also lead to VMID conflicts.

Its suggested that you create a new storage where only the node which you want to separate has access. This can be a new export on your NFS or a new Ceph pool, to name a few examples. Its just important that the exact same storage does not gets accessed by multiple clusters. After setting this storage up move all data from the node and its VMs to it. Then you are ready to separate the node from the cluster.

Warning Ensure all shared resources are cleanly separated! Otherwise you will run into conflicts and problems.

First stop the corosync and the pve-cluster services on the node:

systemctl stop pve-cluster
systemctl stop corosync

Start the cluster filesystem again in local mode:

pmxcfs -l

Delete the corosync configuration files:

rm /etc/pve/corosync.conf
rm /etc/corosync/*

You can now start the filesystem again as normal service:

killall pmxcfs
systemctl start pve-cluster

The node is now separated from the cluster. You can deleted it from a remaining node of the cluster with:

pvecm delnode oldnode

If the command failed, because the remaining node in the cluster lost quorum when the now separate node exited, you may set the expected votes to 1 as a workaround:

pvecm expected 1

And then repeat the pvecm delnode command.

Now switch back to the separated node, here delete all remaining files left from the old cluster. This ensures that the node can be added to another cluster again without problems.

rm /var/lib/corosync/*

As the configuration files from the other nodes are still in the cluster filesystem you may want to clean those up too. Remove simply the whole directory recursive from /etc/pve/nodes/NODENAME, but check three times that you used the correct one before deleting it.

Caution The nodes SSH keys are still in the authorized_key file, this means the nodes can still connect to each other with public key authentication. This should be fixed by removing the respective keys from the /etc/pve/priv/authorized_keys file.

Quorum

Proxmox VE use a quorum-based technique to provide a consistent state among all cluster nodes.

A quorum is the minimum number of votes that a distributed transaction has to obtain in order to be allowed to perform an operation in a distributed system.

Quorum (distributed computing)
— from Wikipedia

In case of network partitioning, state changes requires that a majority of nodes are online. The cluster switches to read-only mode if it loses quorum.

Note Proxmox VE assigns a single vote to each node by default.

Cluster Network

The cluster network is the core of a cluster. All messages sent over it have to be delivered reliably to all nodes in their respective order. In Proxmox VE this part is done by corosync, an implementation of a high performance, low overhead high availability development toolkit. It serves our decentralized configuration file system (pmxcfs).

Network Requirements

This needs a reliable network with latencies under 2 milliseconds (LAN performance) to work properly. The network should not be used heavily by other members, ideally corosync runs on its own network. Do not use a shared network for corosync and storage (except as a potential low-priority fallback in a redundant configuration).

Before setting up a cluster, it is good practice to check if the network is fit for that purpose. To make sure the nodes can connect to each other on the cluster network, you can test the connectivity between them with the ping tool.

If the Proxmox VE firewall is enabled, ACCEPT rules for corosync will automatically be generated - no manual action is required.

Note Corosync used Multicast before version 3.0 (introduced in Proxmox VE 6.0). Modern versions rely on Kronosnet for cluster communication, which, for now, only supports regular UDP unicast.
Caution You can still enable Multicast or legacy unicast by setting your transport to udp or udpu in your corosync.conf, but keep in mind that this will disable all cryptography and redundancy support. This is therefore not recommended.

Separate Cluster Network

When creating a cluster without any parameters the corosync cluster network is generally shared with the Web UI and the VMs and their traffic. Depending on your setup, even storage traffic may get sent over the same network. Its recommended to change that, as corosync is a time critical real time application.

Setting Up A New Network

First you have to set up a new network interface. It should be on a physically separate network. Ensure that your network fulfills the cluster network requirements.

Separate On Cluster Creation

This is possible via the linkX parameters of the pvecm create command used for creating a new cluster.

If you have set up an additional NIC with a static address on 10.10.10.1/25, and want to send and receive all cluster communication over this interface, you would execute:

pvecm create test --link0 10.10.10.1

To check if everything is working properly execute:

systemctl status corosync

Afterwards, proceed as described above to add nodes with a separated cluster network.

Separate After Cluster Creation

You can do this if you have already created a cluster and want to switch its communication to another network, without rebuilding the whole cluster. This change may lead to short durations of quorum loss in the cluster, as nodes have to restart corosync and come up one after the other on the new network.

Check how to edit the corosync.conf file first. Then, open it and you should see a file similar to:

logging {
  debug: off
  to_syslog: yes
}

nodelist {

  node {
    name: due
    nodeid: 2
    quorum_votes: 1
    ring0_addr: due
  }

  node {
    name: tre
    nodeid: 3
    quorum_votes: 1
    ring0_addr: tre
  }

  node {
    name: uno
    nodeid: 1
    quorum_votes: 1
    ring0_addr: uno
  }

}

quorum {
  provider: corosync_votequorum
}

totem {
  cluster_name: testcluster
  config_version: 3
  ip_version: ipv4-6
  secauth: on
  version: 2
  interface {
    linknumber: 0
  }

}
Note ringX_addr actually specifies a corosync link address, the name "ring" is a remnant of older corosync versions that is kept for backwards compatibility.

The first thing you want to do is add the name properties in the node entries if you do not see them already. Those must match the node name.

Then replace all addresses from the ring0_addr properties of all nodes with the new addresses. You may use plain IP addresses or hostnames here. If you use hostnames ensure that they are resolvable from all nodes. (see also Link Address Types)

In this example, we want to switch the cluster communication to the 10.10.10.1/25 network. So we replace all ring0_addr respectively.

Note The exact same procedure can be used to change other ringX_addr values as well, although we recommend to not change multiple addresses at once, to make it easier to recover if something goes wrong.

After we increase the config_version property, the new configuration file should look like:

logging {
  debug: off
  to_syslog: yes
}

nodelist {

  node {
    name: due
    nodeid: 2
    quorum_votes: 1
    ring0_addr: 10.10.10.2
  }

  node {
    name: tre
    nodeid: 3
    quorum_votes: 1
    ring0_addr: 10.10.10.3
  }

  node {
    name: uno
    nodeid: 1
    quorum_votes: 1
    ring0_addr: 10.10.10.1
  }

}

quorum {
  provider: corosync_votequorum
}

totem {
  cluster_name: testcluster
  config_version: 4
  ip_version: ipv4-6
  secauth: on
  version: 2
  interface {
    linknumber: 0
  }

}

Then, after a final check if all changed information is correct, we save it and once again follow the edit corosync.conf file section to bring it into effect.

The changes will be applied live, so restarting corosync is not strictly necessary. If you changed other settings as well, or notice corosync complaining, you can optionally trigger a restart.

On a single node execute:

systemctl restart corosync

Now check if everything is fine:

systemctl status corosync

If corosync runs again correct restart corosync also on all other nodes. They will then join the cluster membership one by one on the new network.

Corosync addresses

A corosync link address (for backwards compatibility denoted by ringX_addr in corosync.conf) can be specified in two ways:

  • IPv4/v6 addresses will be used directly. They are recommended, since they are static and usually not changed carelessly.

  • Hostnames will be resolved using getaddrinfo, which means that per default, IPv6 addresses will be used first, if available (see also man gai.conf). Keep this in mind, especially when upgrading an existing cluster to IPv6.

Caution Hostnames should be used with care, since the address they resolve to can be changed without touching corosync or the node it runs on - which may lead to a situation where an address is changed without thinking about implications for corosync.

A seperate, static hostname specifically for corosync is recommended, if hostnames are preferred. Also, make sure that every node in the cluster can resolve all hostnames correctly.

Since Proxmox VE 5.1, while supported, hostnames will be resolved at the time of entry. Only the resolved IP is then saved to the configuration.

Nodes that joined the cluster on earlier versions likely still use their unresolved hostname in corosync.conf. It might be a good idea to replace them with IPs or a seperate hostname, as mentioned above.

Corosync Redundancy

Corosync supports redundant networking via its integrated kronosnet layer by default (it is not supported on the legacy udp/udpu transports). It can be enabled by specifying more than one link address, either via the --linkX parameters of pvecm (while creating a cluster or adding a new node) or by specifying more than one ringX_addr in corosync.conf.

Note To provide useful failover, every link should be on its own physical network connection.

Links are used according to a priority setting. You can configure this priority by setting knet_link_priority in the corresponding interface section in corosync.conf, or, preferrably, using the priority parameter when creating your cluster with pvecm:

 # pvecm create CLUSTERNAME --link0 10.10.10.1,priority=20 --link1 10.20.20.1,priority=15

This would cause link1 to be used first, since it has the lower priority.

If no priorities are configured manually (or two links have the same priority), links will be used in order of their number, with the lower number having higher priority.

Even if all links are working, only the one with the highest priority will see corosync traffic. Link priorities cannot be mixed, i.e. links with different priorities will not be able to communicate with each other.

Since lower priority links will not see traffic unless all higher priorities have failed, it becomes a useful strategy to specify even networks used for other tasks (VMs, storage, etc…) as low-priority links. If worst comes to worst, a higher-latency or more congested connection might be better than no connection at all.

To add a new link to a running configuration, first check how to edit the corosync.conf file.

Then, add a new ringX_addr to every node in the nodelist section. Make sure that your X is the same for every node you add it to, and that it is unique for each node.

Lastly, add a new interface, as shown below, to your totem section, replacing X with your link number chosen above.

Assuming you added a link with number 1, the new configuration file could look like this:

logging {
  debug: off
  to_syslog: yes
}

nodelist {

  node {
    name: due
    nodeid: 2
    quorum_votes: 1
    ring0_addr: 10.10.10.2
    ring1_addr: 10.20.20.2
  }

  node {
    name: tre
    nodeid: 3
    quorum_votes: 1
    ring0_addr: 10.10.10.3
    ring1_addr: 10.20.20.3
  }

  node {
    name: uno
    nodeid: 1
    quorum_votes: 1
    ring0_addr: 10.10.10.1
    ring1_addr: 10.20.20.1
  }

}

quorum {
  provider: corosync_votequorum
}

totem {
  cluster_name: testcluster
  config_version: 4
  ip_version: ipv4-6
  secauth: on
  version: 2
  interface {
    linknumber: 0
  }
  interface {
    linknumber: 1
  }
}

The new link will be enabled as soon as you follow the last steps to edit the corosync.conf file. A restart should not be necessary. You can check that corosync loaded the new link using:

journalctl -b -u corosync

It might be a good idea to test the new link by temporarily disconnecting the old link on one node and making sure that its status remains online while disconnected:

pvecm status

If you see a healthy cluster state, it means that your new link is being used.

Corosync External Vote Support

This section describes a way to deploy an external voter in a Proxmox VE cluster. When configured, the cluster can sustain more node failures without violating safety properties of the cluster communication.

For this to work there are two services involved:

  • a so called qdevice daemon which runs on each Proxmox VE node

  • an external vote daemon which runs on an independent server.

As a result you can achieve higher availability even in smaller setups (for example 2+1 nodes).

QDevice Technical Overview

The Corosync Quroum Device (QDevice) is a daemon which runs on each cluster node. It provides a configured number of votes to the clusters quorum subsystem based on an external running third-party arbitrator’s decision. Its primary use is to allow a cluster to sustain more node failures than standard quorum rules allow. This can be done safely as the external device can see all nodes and thus choose only one set of nodes to give its vote. This will only be done if said set of nodes can have quorum (again) when receiving the third-party vote.

Currently only QDevice Net is supported as a third-party arbitrator. It is a daemon which provides a vote to a cluster partition if it can reach the partition members over the network. It will give only votes to one partition of a cluster at any time. It’s designed to support multiple clusters and is almost configuration and state free. New clusters are handled dynamically and no configuration file is needed on the host running a QDevice.

The external host has the only requirement that it needs network access to the cluster and a corosync-qnetd package available. We provide such a package for Debian based hosts, other Linux distributions should also have a package available through their respective package manager.

Note In contrast to corosync itself, a QDevice connects to the cluster over TCP/IP. The daemon may even run outside of the clusters LAN and can have longer latencies than 2 ms.

Supported Setups

We support QDevices for clusters with an even number of nodes and recommend it for 2 node clusters, if they should provide higher availability. For clusters with an odd node count we discourage the use of QDevices currently. The reason for this, is the difference of the votes the QDevice provides for each cluster type. Even numbered clusters get single additional vote, with this we can only increase availability, i.e. if the QDevice itself fails we are in the same situation as with no QDevice at all.

Now, with an odd numbered cluster size the QDevice provides (N-1) votes — where N corresponds to the cluster node count. This difference makes sense, if we had only one additional vote the cluster can get into a split brain situation. This algorithm would allow that all nodes but one (and naturally the QDevice itself) could fail. There are two drawbacks with this:

  • If the QNet daemon itself fails, no other node may fail or the cluster immediately loses quorum. For example, in a cluster with 15 nodes 7 could fail before the cluster becomes inquorate. But, if a QDevice is configured here and said QDevice fails itself no single node of the 15 may fail. The QDevice acts almost as a single point of failure in this case.

  • The fact that all but one node plus QDevice may fail sound promising at first, but this may result in a mass recovery of HA services that would overload the single node left. Also ceph server will stop to provide services after only ((N-1)/2) nodes are online.

If you understand the drawbacks and implications you can decide yourself if you should use this technology in an odd numbered cluster setup.

QDevice-Net Setup

We recommend to run any daemon which provides votes to corosync-qdevice as an unprivileged user. Proxmox VE and Debian provides a package which is already configured to do so. The traffic between the daemon and the cluster must be encrypted to ensure a safe and secure QDevice integration in Proxmox VE.

First install the corosync-qnetd package on your external server and the corosync-qdevice package on all cluster nodes.

After that, ensure that all your nodes on the cluster are online.

You can now easily set up your QDevice by running the following command on one of the Proxmox VE nodes:

pve# pvecm qdevice setup <QDEVICE-IP>

The SSH key from the cluster will be automatically copied to the QDevice. You might need to enter an SSH password during this step.

After you enter the password and all the steps are successfully completed, you will see "Done". You can check the status now:

pve# pvecm status

...

Votequorum information
~~~~~~~~~~~~~~~~~~~~~
Expected votes:   3
Highest expected: 3
Total votes:      3
Quorum:           2
Flags:            Quorate Qdevice

Membership information
~~~~~~~~~~~~~~~~~~~~~~
    Nodeid      Votes    Qdevice Name
    0x00000001          1    A,V,NMW 192.168.22.180 (local)
    0x00000002          1    A,V,NMW 192.168.22.181
    0x00000000          1            Qdevice

which means the QDevice is set up.

Frequently Asked Questions

Tie Breaking

In case of a tie, where two same-sized cluster partitions cannot see each other but the QDevice, the QDevice chooses randomly one of those partitions and provides a vote to it.

Possible Negative Implications

For clusters with an even node count there are no negative implications when setting up a QDevice. If it fails to work, you are as good as without QDevice at all.

Adding/Deleting Nodes After QDevice Setup

If you want to add a new node or remove an existing one from a cluster with a QDevice setup, you need to remove the QDevice first. After that, you can add or remove nodes normally. Once you have a cluster with an even node count again, you can set up the QDevice again as described above.

Removing the QDevice

If you used the official pvecm tool to add the QDevice, you can remove it trivially by running:

pve# pvecm qdevice remove

Corosync Configuration

The /etc/pve/corosync.conf file plays a central role in a Proxmox VE cluster. It controls the cluster membership and its network. For further information about it, check the corosync.conf man page:

man corosync.conf

For node membership you should always use the pvecm tool provided by Proxmox VE. You may have to edit the configuration file manually for other changes. Here are a few best practice tips for doing this.

Edit corosync.conf

Editing the corosync.conf file is not always very straightforward. There are two on each cluster node, one in /etc/pve/corosync.conf and the other in /etc/corosync/corosync.conf. Editing the one in our cluster file system will propagate the changes to the local one, but not vice versa.

The configuration will get updated automatically as soon as the file changes. This means changes which can be integrated in a running corosync will take effect immediately. So you should always make a copy and edit that instead, to avoid triggering some unwanted changes by an in-between safe.

cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new

Then open the config file with your favorite editor, nano and vim.tiny are preinstalled on any Proxmox VE node for example.

Note Always increment the config_version number on configuration changes, omitting this can lead to problems.

After making the necessary changes create another copy of the current working configuration file. This serves as a backup if the new configuration fails to apply or makes problems in other ways.

cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak

Then move the new configuration file over the old one:

mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf

You may check with the commands

systemctl status corosync
journalctl -b -u corosync

If the change could be applied automatically. If not you may have to restart the corosync service via:

systemctl restart corosync

On errors check the troubleshooting section below.

Troubleshooting

Issue: quorum.expected_votes must be configured

When corosync starts to fail and you get the following message in the system log:

[...]
corosync[1647]:  [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
corosync[1647]:  [SERV  ] Service engine 'corosync_quorum' failed to load for reason
    'configuration error: nodelist or quorum.expected_votes must be configured!'
[...]

It means that the hostname you set for corosync ringX_addr in the configuration could not be resolved.

Write Configuration When Not Quorate

If you need to change /etc/pve/corosync.conf on an node with no quorum, and you know what you do, use:

pvecm expected 1

This sets the expected vote count to 1 and makes the cluster quorate. You can now fix your configuration, or revert it back to the last working backup.

This is not enough if corosync cannot start anymore. Here its best to edit the local copy of the corosync configuration in /etc/corosync/corosync.conf so that corosync can start again. Ensure that on all nodes this configuration has the same content to avoid split brains. If you are not sure what went wrong it’s best to ask the Proxmox Community to help you.

Corosync Configuration Glossary

ringX_addr

This names the different link addresses for the kronosnet connections between nodes.

Cluster Cold Start

It is obvious that a cluster is not quorate when all nodes are offline. This is a common case after a power failure.

Note It is always a good idea to use an uninterruptible power supply (“UPS”, also called “battery backup”) to avoid this state, especially if you want HA.

On node startup, the pve-guests service is started and waits for quorum. Once quorate, it starts all guests which have the onboot flag set.

When you turn on nodes, or when power comes back after power failure, it is likely that some nodes boots faster than others. Please keep in mind that guest startup is delayed until you reach quorum.

Guest Migration

Migrating virtual guests to other nodes is a useful feature in a cluster. There are settings to control the behavior of such migrations. This can be done via the configuration file datacenter.cfg or for a specific migration via API or command line parameters.

It makes a difference if a Guest is online or offline, or if it has local resources (like a local disk).

For Details about Virtual Machine Migration see the QEMU/KVM Migration Chapter.

For Details about Container Migration see the Container Migration Chapter.

Migration Type

The migration type defines if the migration data should be sent over an encrypted (secure) channel or an unencrypted (insecure) one. Setting the migration type to insecure means that the RAM content of a virtual guest gets also transferred unencrypted, which can lead to information disclosure of critical data from inside the guest (for example passwords or encryption keys).

Therefore, we strongly recommend using the secure channel if you do not have full control over the network and can not guarantee that no one is eavesdropping to it.

Note Storage migration does not follow this setting. Currently, it always sends the storage content over a secure channel.

Encryption requires a lot of computing power, so this setting is often changed to "unsafe" to achieve better performance. The impact on modern systems is lower because they implement AES encryption in hardware. The performance impact is particularly evident in fast networks where you can transfer 10 Gbps or more.

Migration Network

By default, Proxmox VE uses the network in which cluster communication takes place to send the migration traffic. This is not optimal because sensitive cluster traffic can be disrupted and this network may not have the best bandwidth available on the node.

Setting the migration network parameter allows the use of a dedicated network for the entire migration traffic. In addition to the memory, this also affects the storage traffic for offline migrations.

The migration network is set as a network in the CIDR notation. This has the advantage that you do not have to set individual IP addresses for each node. Proxmox VE can determine the real address on the destination node from the network specified in the CIDR form. To enable this, the network must be specified so that each node has one, but only one IP in the respective network.

Example

We assume that we have a three-node setup with three separate networks. One for public communication with the Internet, one for cluster communication and a very fast one, which we want to use as a dedicated network for migration.

A network configuration for such a setup might look as follows:

iface eno1 inet manual

# public network
auto vmbr0
iface vmbr0 inet static
    address 192.X.Y.57
    netmask 255.255.250.0
    gateway 192.X.Y.1
    bridge_ports eno1
    bridge_stp off
    bridge_fd 0

# cluster network
auto eno2
iface eno2 inet static
    address  10.1.1.1
    netmask  255.255.255.0

# fast network
auto eno3
iface eno3 inet static
    address  10.1.2.1
    netmask  255.255.255.0

Here, we will use the network 10.1.2.0/24 as a migration network. For a single migration, you can do this using the migration_network parameter of the command line tool:

# qm migrate 106 tre --online --migration_network 10.1.2.0/24

To configure this as the default network for all migrations in the cluster, set the migration property of the /etc/pve/datacenter.cfg file:

# use dedicated migration network
migration: secure,network=10.1.2.0/24
Note The migration type must always be set when the migration network gets set in /etc/pve/datacenter.cfg.