We use an EMC SAN appliance. There are 2 'paths' from the cluster to the SAN. Today we experienced that one of the LUN paths was 'dead' (as reported in the windows event logs). Presumably the multipath LUN was down to a single path. At this time we saw I/O errors on the SQL Server that "I/O requests taking longer than 15 seconds to complete...". After we failed to the passive node in the cluster, the problem went away.
QUESTION: If the primary cluster node could no longer communicate along one of the 2 configured paths -- does that effectively mean that bandwidth to the LUN was split in half?? Also, could a job that used that SAN drive heavily for reads/writes (our TempDB is on this drive) cause a LUN path to appear 'down'?
0 Answers