HACMP Configuration Q's

[octar]

Hi,

I have a cascading cluster environment:
With using printers, I want to know if on failover the printer queues will be taken over or do I have to define them on the system or in a script?
Also I have also addedd nfs connections and updated the /etc/exports file along with my scripts, do I need to copy this file to the takeover node or is that executed in a script?
ps. How do I check what is executed on takeover? scripts??
thanks

 

[aixat]

do search on hacmp and redbook on this forum you will find the answer.

 

[StevenLangdale]

Octar

I use HACMP/ES 4.4, therefore all my answers will be pertinent to that. Having said that, I'm sure this is the same on all versions, shout out if not!

Print queues will not be taken over, you will need to create them on all participating nodes in the cluster. Also, unless you make an new highly available filesystem and mount is over the spool directory, you will loose any pending print jobs on failover too.

Thanks

Steven

 

[DSMARWAY]

hi,
printers you need to add to both servers , becuase if server a crashes where will the jobs get queued currently they are under /var/adm/qdaemon . BAsically what Steven said?

The nfs , as long as you have created your nfs in the shared volume group and in the cluster resources you have defined in the filesystems/directories to export
you should be .o.k

but note :- you must create filesystems, nfs etc on node1 and then import them on node 2 once that has been done
you just add the necessary info in the cluster resources for your resource group.

Then do a synchronise resources ( emulate it first ) if no errors then do an actual this will make sure node 2 has same config.


Check /tmp/hacmp.out file to find out exactly what is being executed when doing a takeover. The log is very detailed and you may need to look at it 50 times or more to understand what it is running , thats IBM for you .
Basically you have promary event scripts and secondary event scripts that run depending on what is happening in your clusterfor e.g. fail_standby is a primary event script which runs when your standby card has failed
you can view all the evnts i. cluster resources -- events


HTH

 

[stubbint]

You need to create the spool directories on a filesystem that fails over, and have them mount over those on the rootvg.
You also need to keep /etc/qconfig in sync on all the cluster nodes. I suggest making /etc/qconfig a link to a qconfig file located on the above failing over filesystem.
Create start_printers and stop_printers scripts that perform "startsrc -g spooler" or "stopsrc -g spooler" commands when the nodes acquires or releases the resources.