vgchange: "Activation mode requested for the volume group conflicts with configured mode"
Two node Serviceguard eRAC cluster with several shared volume groups activated in a shared server and shared client environment.
After rebooting node_1 some volume groups cannot be activated any longer.
node_1 # vgchange -a s /dev/vg14vgchange: Activation mode requested for the volume group "/dev/vg14" conflicts with configured mode.
All the volume groups are ok on node_2 and the VG Status is available, shared, server.
User was able to activate the shared VG’s with vgchange -a y on node_1, which should not happen.
The volume groups then were deactivated with vgchange -a n.
This issue matches QXCR1001161003 with cluster-ID corruption.
A workaround is available to have the node which has the VG’s activated in shared mode to post a new lvmrec with the activation mode for shared. Then, vgchange -a s activation on the other node will work.
Workaround below is performed on volume group vg14:
- Change IO timeout temporarily, with pvchange -t, which forces node_2 to write the LVMREC on each PV in vg14.
- For each PV of vg14, check its original "IO timeout" value with pvdisplay.If the value is shown as "default", it means that the IO timeout value is set to 0.Example:node_2> # pvdisplay /dev/dsk/cXtYdZ --- Physical volumes ---PV Name /dev/dsk/c4t13d0VG Name /dev/ p1048vg14PV Status available...IO Timeout (Seconds) defaultAutoswitch OnProactive Polling On
- Change the IO timeout value to a different value, say 30 seconds, on each PV.This requires exclusive activation on the volume group so the volume group needs to be cross-activated in exclusive mode on node_2 first.Example:node_2># pvchange -t 30 /dev/dsk/cXtYdZpvchange: The volume group "/dev/ vg14" is active in Shared Mode.Cannot perform configuration change.node_2># vgchange -a e -x /dev/vg14Volume group "/dev/vg14" has been successfully changed.node_2># pvchange -t 30 /dev/dsk/cXtYdZPhysical volume "/dev/dsk/cXtYdZ" has been successfully changed.Volume Group configuration for /dev/vg14 has been saved in /etc/lvmconf/vg14.conf
- The cluster ID should have been re-instated on each PV at this point.
- Now, restore the IO timeout value and switch back to shared-mode activationnode_2># pvchange -t 0 /dev/dsk/cXtYdZPhysical volume "/dev/dsk/cXtYdZ" has been successfully changed.Volume Group configuration for /dev/vg14 has been saved in /etc/lvmconf/vg14.confsjhsok2># vgchange -a s -x vg14Volume group "/dev/vg14" has been successfully changed.
- On node_1, now the user should be able to activate the volume group in shared mode.Example:node_1># vgchange -a s /dev/vg14Activated volume group in Shared Mode.This node is a Client.Volume group "/dev/vg14" has been successfully changed.