service guard

[Kgreen]

I have to apply a patch to my cluster. Now have about 14 clusters and the requirement for this patch is that the cluster should be down completely along with the nodes. Does anyone has a script that would stop all the packages , stop the nodes.. and do basic check if the cluster is completely halted

[Calandrello]

Friend executes cmhaltcl - f executes cmviewcl this must show, all the packages and nodes as down. to initiate everything it executes cmruncl - f again

[Doug O'Leary]

Hey;

No script necessary. To reiterate, and hopefully clarify the previous response:

cmhaltcl -f

executed on any node of the cluster will halt all the packages and then halt the cluster.

Once you get your command prompt back, you can issue

cmviewcl

to verify the cluster is down. Apply your patch on all nodes, then issue:

cmruncl

on one and only one node of each cluster.

HTH;

Doug

[IT_2007]

It is always recommended to check status of cluster while issuing cmhaltcl -f command since packages may encounter problems like disk failures, hardware failures etc..

If you encounter problem during shutdown how do you know/fix problems unless you visit package logs, syslogs?

If clusters were up and running for long time then better take one cluster at a time and apply patch. Remeber that criticality of applucations, databases, shut them down before you bring down cluster.

[Kgreen]

Thanks guys I know that cmhaltcl would do the job. But I was looking in more for like find all the package running on that node and shutit down, and then do a cmhaltnode. These are some new configs and was not able to check all things on this duw to lack of time. anyway I have scripted things , was just wondering if something better was available...