MetroCluster and VxVM

[Milan Mazac]

When I start cluster services on primary node the cluster try to do clearimport of all VxVM diskgroups. Failover node doesn't. It could be seen in /etc/rc.log file. It is problem of failback bacause VxVM has disabled disk pathes. It can be rocevered by vxdctl enable but is only workaround.

[Albert_31]

Hello Milan,

a) Does it happen always or you faced it only once..

b) Does the failover node join the cluster group.

c) Why do you need to run the vxdctl enable ?
regards

Albert

[Milan Mazac]

Hello Albert,

ad a) It happen always
ad b) Yes, the failover node is a member of cluster

ad c) according my investigation after cluster startup the {as a reaction on vxdg clearimport) all vxvm disks have disabled their pathes to the physical devices. After vxdctl enable vxvm make a rescan of devices and pathes became enabled.

Milan

[Albert_31]

Hello Milan,

Sorry for the very late reply, forgot to enable notification for this thread...regarding the vxvm issue

c) according my investigation after cluster startup the {as a reaction on vxdg clearimport) all vxvm disks have disabled their pathes to the physical devices. After vxdctl enable vxvm make a rescan of devices and pathes became enabled.

vxdg clearimport only removes any import locks on the diskgroup it should not disable the paths..

To further isolate the issue..
a) what is the storage used here (common)
b) are you using any load balancing software like powerpath,securepath..vxvm dmp etc..
c) can you send me the following from both the nodes

1) ioscan -fn
2) vxdisk list -o alldgs
3) vxdisk list
4) vxprint
5) # vxdmpadm listctlr all

regards

albert
EDited to comply with ITRC Guidelines