Package problem .

[chapaya_1]

Hi ,
I halted a package on node1 and try to run it on the second node , and i got this errors:

Apr 27 13:42:41 - "node2": Activating volume group vg03 with exclusive option.
vgchange: Activation of volume group "/dev/vg03" denied by another node in the cluster.
Request on this system conflicts with Activation Mode on remote system.
ERROR: Function activate_volume_group
ERROR: Failed to activate vg03
Apr 27 13:42:42 - Node "node2": Deactivating volume group vg03
vgchange: Volume group "vg03" has been successfully changed.

[Patrick Wallek]

It sounds like your VG was not deactivated on node1 and thus couldn't be activated on node2. The VG should be set up so that it is "shareable" for the cluster (man vgchange and have a look at the -c and -s options).

[Marvin Strong]

Is vg03 deactivated on node1?
Are you sure the cmhaltpkg was successful?
Or was it just partially successful.

[Lorenzo Facello]

Hi,
try on node2
# vgchange -c y vg03 --> abilitate vg on the cluster
# cmrunpkg pkg_name
Hope this helps
Regds
L.

[Geoff Wild]

Yes - as others have said, it wasn't deactivated on the other node....check the package logfile on the other node as well as syslog....


Rgds...Geoff

[chapaya_1]

Hi ,
Cmhaltpkg run successfully so i don't understand why vg03 wasn't deactivate by node1 .

Bye.

[Marvin Strong]

look through the package logfile, as suggested it should tell you why it wasn't deactivated.

[Sanjay_6 · Accepted solution]

Hi,

Do a "vgdisplay -v" on node1 and see if the VG is active over there. If so deactivate it using "vgchange -a n vg_name" and then try to restart the package on node2.

The log file under /etc/cmcluster/pkg_name/pkg_name.sh.log can give you more info on why the VG did not get deactivated.

Also revisit the package configuration on both the nodes to make sure the package is configured properly.

Hope this helps.

Regds

[SS_6]

In additions to above try to do the following on node one where you have stopped the package.
Halt the whole cluster if you have only package.
#cmhaltcl
On node second
#cmrunnode -n nodename
#cmrunpkg -n nodename -v packageName
Look in cluster log file for any errors.

[generic_1]

In addition I would go back and check your ServiceGuard scripts and make sure it is gracefully deactivating those volume groups. The last thing you want to happen is your failover to fail when you really need it.

[Bharat Katkar]

See how i do it:

#cmhaltpkg
#cmrunpkg -v -n node2
#cmmodpkg -e
#cmviewcl

Check whether your package is running?

And now that you say your VG is not getting deactivated i suspect that some other LV's which are been used by users exists in the VG. LV may not be in cluster or used by some other package as well.

[Rita C Workman]

..hmm...is it possible you made a change to the package for something and didn't complete all the steps (successfully) on your remote node.

Go back and ensure that everything is down on node1. And redo all the steps, as if you just made a change to your pkg...from new mapfiles to vgexport/vgimports & copying over the control files.

Then see if things go up and down cleanly on both.

Rgrds,
Rita

[chapaya_1]

Hi All ,
It looks like a problem with deactivation although the package was shutdown successfully .

Bye.