Hi,
I have setup a Mirrored 2008 SQL pair with an independant witness service.
With a complete hardware failure on the primary SQL server the failover
operates with complete success.

However in the event of the primary SQL server having an issue with the
iscsi volume that the databases and logs are living on the server does not
recover gracefully and most, if not all, of the databases become unavailable.
The mirror does not then fail over.

My question is:

For a SQL 2008 server using iscsi, is there a setting within SQL that I can
set so that in the event of a short (i.e. less than 60 seconds) failure SQL
will cache to memory and wait for the iscsi volume to become available again?
or can I configure the witness server to failover in the event of any issue
being reported to a database?

At the moment the only solution I can think of is to monitor the event log
for the iscsi error and stop SQL service until the iscsi error has been
corrected.

Thanks,
Phil.