|
Qureying the state of a cluster |
|
|
|
Written by Frank Schalude
|
|
Tuesday, 23 March 2010 |
HACMP/PowerHA offers several possibilities to query the state of the cluster. The best way to see the state of each node ist to execute the command lssrc -ls clstrmgrES and to look for the output of the first line for the state of each node. This will ensure that you do not start any action as long as the cluster is not stable (ST_STABLE).
The following short alias is a nice help to quickly get the state of all cluster nodes without the need to login to each node:
alias checkcl='/usr/es/sbin/cluster/cspoc/cli_on_cluster -P "lssrc -ls clstrmgrES|head -n1"'
Be aware that forced down nodes will show off as "ST_STABLE"
The state can be one out of the following:
ST_NOT_CONFIGURED HACMP has never been started on this node
ST_INIT HACMP has at least started one time on this node but is currently not active (e.g. after a reboot)
ST_STABLE HACMP is up and stable
ST_JOINING HACMP is about to start and the node is joining the cluster
ST_VOTING HACMP is joining the cluster and group services are voting
ST_BARRIER The node waits for the completion of an event running on any other node
ST_CBARRIER The node has received the completion of an event of any node
ST_RP_RUNNING The node is currently running alocal event
ST_RP_FAILED The last event failed => check /var/adm/cluster.log for errors.
|
