Error connecting to a cluster node in ispmanager 6 business

Symptoms

Operations involving cluster nodes, e.g. backup and user import, can not be executed.

Checking the SSH connection in the panel section "Cluster nodes" - "Connection" produces an error message.

Failed connection attempts are registered in the main log of the panel:

Apr 18 04:03:56 [4177478:1] backup WARNING SSH connection failed. Attempt: 1 of 3
Apr 18 04:03:56 [4177478:1] backup EXTINFO Wait 2 seconds before next attempt
Apr 18 04:03:58 [4177478:1] proc EXTINFO Run 'ssh -q -o StrictHostKeyChecking=no -p 22 -i etc/.ssh/master_id 123.123.123.123 cd\ /usr/local/mgr5\;\ /bin/sh' pid 4182889
Apr 18 04:03:58 [4177478:1] proc EXTINFO Process 4182889 finished with status 255
Apr 18 04:03:58 [4177478:1] backup WARNING SSH connection failed. Attempt: 2 of 3
Apr 18 04:03:58 [4177478:1] backup EXTINFO Wait 2 seconds before next attempt
Apr 18 04:04:00 [4177478:1] proc EXTINFO Run 'ssh -q -o StrictHostKeyChecking=no -p 22 -i etc/.ssh/master_id 123.123.123.123 cd\ /usr/local/mgr5\;\ /bin/sh' pid 4182984
Apr 18 04:04:00 [4177478:1] proc EXTINFO Process 4182984 finished with status 255
Apr 18 04:04:00 [4177478:1] backup WARNING SSH connection failed. Attempt: 3 of 3

Causes

Invalid SSH key to access cluster node
SSH access is blocked on a remote server (e.g. due to network restrictions or firewall rules)
Network connection issues between servers

Solution

SSH key adjustment

On a cluster node, the /root/.ssh/authorized_keys file must contain a public key from the /usr/local/mgr5/etc/.ssh/master_id.pub file on a master node.

If the public key is missing, add it manually.

If the previous public key or key file has been lost, generate a new key pair - public (/usr/local/mgr5/etc/.ssh/master_id.pub) and private (/usr/local/mgr5/etc/.ssh/master_id).

The same is true for operations performed on the master node, i.e. the master node must be able to connect to itself using the SSH service key specified above.

Network connection check

Check firewall settings on the cluster node and make sure that SSH access is allowed from the master server's side.

Check network connection between the servers to detect possible network issues using "telnet", "nmap" or "traceroute" utilities:

telnet <ip-address> <port>
nmap <ip-address> -p <port>
traceroute <ip-address>

About commands
<ip-address> - the cluster node IP address;
<port> - he port used for SSH (number 22 by default).

Additional

Make sure that the SSHD service settings enable SSH key authorization.

Also, try increasing the cluster node request timeouts.

ispmanager 6