group failed

[linuly]

hi,

when vcs tries to bring up a group on a node and a resource is FAULTED then the group will fails over to other node.If here the resource is Faulted then the group goes OFFLINE on both nodes(suppose we have 2 node cluster).right?

if a diskgroup is brought online by vcs on a node then if i try to import this diskgroup on ther node the DG will be imported but no volume will be seen here.this is because scsi-3 PR.right?

[snockles · Accepted solution]

First part, yes, the group stays offline. The cure is that you fix whatever is causing the fault, and then clear the FAULTED status (hares -clear/hagrp -clear for non-persistent resources, persistent resources will come ONLINE when the fault is fixed). You can then bring the group online.

VCS always plays safe. If a fault has occurred, the administrator has to intervene to deal with the issue.

Second part, I/O fencing is all about protecting your data. It uses a SCSI-3 Reservation (lock) called WriteExclusiveRegistrantsOnly. In a failover environment, when VCS imports a diskgroup, the node registers with the data disks, and puts the reservation in place. Other nodes will still be able to see the disks, but, if you forcibly import on another node, the reservation will ensure you can't write, even if you can see the volumes/data.