topic Re: syslog message in Operating System - HP-UX

syslog message

Javier Ortiz Guajardo — Tue, 18 Mar 2003 15:37:47 GMT

Hi, i??ve receive this message in syslog
cmcld[2463]: Node xxxxxx has ceased cluster activities.
cmcld[2463]: Attempting to form a new cluster
cmcld[2463]: Turning off safety time protection since the
cluster
cmcld[2463]: now consists of a single node. If ServiceGu
ard
cmcld[2463]: fails, this node will not automatically halt
cmcld[2463]: Clearing Cluster Lock
cmcld[2463]: 1 nodes have formed a new cluster, sequence
#44
cmcld[2463]: The new active cluster membership is: xxxxxx
(id=1)
cmcld[2463]: New node xxxxxx is joining the cluster
cmcld[2463]: Attempting to adjust cluster membership
cmcld[2463]: Enabling safety time protection
cmcld[2463]: Clearing Cluster Lock
cmcld[2463]: 2 nodes have formed a new cluster, sequence
#45
cmcld[2463]: The new active cluster membership is: xxxxxx
(id=1), xxxxxx(id=2)
mvi902 cmcld[2463]: Request from node xxxxxx to disable package
dbnosap on node xxxxxx.
cmcld[2463]: Disabled package dbnosap on node xxxxxx.

do you why this happen and what is the solution to this message.

Thanks.

Re: syslog message

Marco Santerre — Tue, 18 Mar 2003 15:43:19 GMT

Well, from what I understand here, it looks like something happened between your two nodes in your cluster. I figure that if you are posting this here, chances are the second node of your cluster has not actually had a problem (like rebooting or something like that).

Basically, what it did here, is that it reformed your cluster with only one node out of the two.

My gut feeling tells me that something happened with the heartbeat or with one of the two nodes. It could be network related, or anything similar. If you lose you rheartbeat, though, your cluster will re-formed itself with only one of the nodes, whichever ones catches the cluster lock disk faster.

Hope this helps a bit.

Re: syslog message

Michael Steele_2 — Tue, 18 Mar 2003 15:50:34 GMT

I'd put in new SG bininaries and assume the old ones got corrupted, as they occaisionally do.

You can evaluate your existing binaries with :

# cmgetconf -C cltemp.ascii (* cluster binary *)

then run

cmcheckconf -C cltemp.ascii

Re: syslog message

keith persons — Tue, 18 Mar 2003 18:46:05 GMT

before actually redistributing the binary might want to check the node timeout variable in the ascii cluster config file. The default is 2 seconds - if you're running on a switched network, 8 seconds is the recommended minimum to prevent the hb timeouts. If you're getting these message repeatedly in the syslog file than I would expect the node timeout to be too small.

Keith