unable to run the package in alternate node

[rajendra857]

gavnh211:/home/di3850>/usr/sbin/cmviewcl

CLUSTER         STATUS
GT_PR_APP_CL2   up

  NODE           STATUS       STATE
  gavzh211       up           running

    PACKAGE        STATUS           STATE            AUTO_RUN    NODE
    gtaph231       up               running          enabled     gavzh211

  NODE           STATUS       STATE
  gavnh211       up           running

    PACKAGE        STATUS           STATE            AUTO_RUN    NODE
    gtaph233       up               running          enabled     gavnh211
gavnh211:/home/di3850>

when i shutdown the machine gavnh211 the package didnot failover but AUTO_RUN is enabled for the package in gavnh211.

when i tried to run the package gtaph233 in gavzh211 it failed to start and the error mesgae i got is below.

Rajendra P Man...    root@gavnh211:/# cmrunpkg -n gavzh211 gtaph233
Running package gtaph233 on node gavzh211
The package script for gtaph233 failed with no restart. gtaph233 should not be restarted
Unable to run package gtaph233 on node gavzh211
Check the syslog and pkg log files for more detailed information
cmrunpkg: Unable to start some package or package instances.

i saw one error message in gavnh211 pkg.cntl.log file

Sep 18 18:23:09 - Node "gavnh211": Unmounting filesystem on /dev/vgprita/lvprita
umount: cannot unmount /dev/vgprita/lvprita : Device busy
umount: return error 1.
        WARNING:   Running fuser to remove anyone using the file system directly.
/dev/vgprita/lvprita:     8019co(itaoper1)

Sep 18 18:23:09 - Node "gavnh211": Deactivating volume group vgprita
Deactivated volume group in Exclusive Mode.
Volume group "vgprita" has been successfully changed.

i am suspecting the vg not activated on gavzh211 and the problem with package script.It showed exclusive mode.this is a failover cluster.

so i think cluster needs to be activated in cluserawaremode.

kindly suggest and i am attaching the sginfo from one node

[Ralf_S]

Hy,

My first idea; any of the VGs of the packet is still marked as activates in the primary node. may be some LV is mounted there, too.

In general,

check the logfile for the clusterpackage, for the cluster and syslog (in this order) on the system, wher the package has been shut down. I suppose, there are some ERROR or FAILED entiries.

Check the logfile of the package on the server, where you tried to start the package, too.

Now, your cluster is in a state, where the package can not be started on it's alternate node.

To solve this, please type

cmmodpkg -e <servername>

It is important, that you specivy the servername. When you got the message "... failed with no restart" 

The simple command cmmodpkg -e , without servername will not suffice in this situation. It will not reenable the package on all servernodes.

Reading all yout text, yes, you have filesystemproblems. Check all mounts, LVs and VGs, belonging to your package. Umount and vgchange all as required.

To solve the problem completely, go to your package control file (if you have it, otherwise to your config file)

You may add some sleep command before or after the umount command to give the umount and the act of stopping all processes more time to finish.

Alternatively or additionaly, you can / should) set the the following values higher as they are now:

DEACTIVATION_RETRY_COUNT=

FS_UMOUNT_COUNT=

They define, how often deactivation or umount is executed, if the comand is not successfull in the first place.

The sleep from above would be executed in between those moultple retries .

I hope, this is of some help for you.

Bye

Ralf