To prevent disconnections on client side in your SSH terminal you can add this parameter to the configuration file /etc/ssh/ssh_config:
Basically this option will send a null packet to the remote host every 100 seconds and will keep alive the connection.
If you get this message on logs:
VFR-4-FRAG_TABLE_OVERFLOW: Dialer0: the fragment table has reached its maximum threshold 16
usually indicates than you are receiving a high number of fragmented packets. It may happens using p2p applications like torrent or emule, under an fragmentation packet attack or another causes.
You can increase this value and check if warnings disappear:
router(config-if)#ip virtual-reassembly max-reassemblies 64
On occasion it may be useful, erase the entire CIB on pacemaker, for example if you have made a lot of changes, or tested several configurations with errors. Deleting entire CIB data can be a good option for starting again from the scratch.
The command is:
~ $ cibadmin -E --force
Be careful because this command will delete all cluster data, it’s recommended make a backup of the cluster configuration before:
~ $ crm configure show > cib.backup
In pacemaker you can stop a clone resource on one node using the command:
~ $ crm_resource --ban --resource resource1 --host server01
This bans the resource at the specified node, to revert it:
~ $ crm_resource --ban --resource resource1 --host server01 --clear
So you can make system updates, managements or any maintenance task.