Just want to mention that there are a lot of layers
between SQL Server and the eventual disk media, and things
can definitely go wrong at any of these layers.

A while back, we had a nasty problem with SAN disks
disappearing momentarily or SQL Server having problem
writing to a SAN disk. This caused a lot of people much
lost sleep. Because it wasn't clear where exactly the root
of the problem lies, it was a pretty horrific experience
getting multiple vendors including Microsoft, Veritas,
Emulex, Brocade, and EMC all together to troubleshoot the
problem. Eventually, it turned out to be an issue with the
Veritas Volume Manager.

Linchi

i used the /3GB and /PAE Switch on my SQL 2000 cluster a while back and got
the same errors as you did with regard to Insufficient system resources.
Once I stopped using those switches the problem was resolved. I was noticing
that I was running high for nonpaged memory pool. There are tons of
documents about these switches and their effects on system memory and PTE's.
My guess is if you remove those switches your cluster will be fine. You will
have to spend a lot of time tweaking memory settings in the registry in
order to not run out of resources using these switches. It's too risky and
can and a bad mem config can blue screen your cluster. I would first try
just using the /3GB switch to give sql some more mem to play with but would
avoid the /PAE and using AWE with SQL. My guess is that these switches are
pot luck and can work for some systems and need a lot on configuration
testing on others... Good Luck.

"SQL DBA" <anonymous (AT) discussions (DOT) microsoft.com> wrote