Can some one explain me, when the path to a LUN is made unavailable (for example by doing port disable on an FC switch) how the user space and kernel space status fields are getting updated? Which is the event generator for notifying the path fail and who is waiting for the event?
When the path is made available again, the kernel space status is not getting updated properly. The following is the states of the path.
When path is made unavailable: status of active path is [faulty][active]
When IO is made on the DM device, it continues on the other path : status of previous active path becomes [faulty][failed]
When the path is made available: status of the path becomes [ready][failed] from [faulty][failed].
Is this the expected behaviour or should the path status come back to [ready][active] state ?
When the multipath tool is manually exec’ed, the status of path becomes [ready][active]. I tried reducing the path checking interval to lesser value (5 seconds), but the path status remained the same. I am seeing the same behaviour with the latest multipath tools (multipath-tools-0.4.5-pre2.tar.gz).
Thanks and regards