Every few weeks our NW6.5sp3 cluster server goes down to a crawl.

The CPU util is not always high.

What is unique the the Current Disk Requests remaining very high (between
100 - 300 constant).
Our problem is always on the cluster node that holds the user's P: drive.

This P: drive is used for Lotus Notes 6.

Problem is usually first thing in the morning when the users get on - it
happens for about 30 mins or so.

Most days everything hums normally - but every few weeks - down to a crawl.
When things are slow, you can open windows explorer, click on the p: drive
and just wait and wait until seeing any files. (usually it's instant).

It looks like the storage system can't keep up with the requests.

We have 2gb fibre QL2300 going to a MSA1000 with 500gb volumes (SCSI drives
are RAID5 in SAN).

Strangely enough, we have narrowed it down to a particular volume on the
SAN. It appears if it's the SAN that is slow - but the other volumes on the
same SAN are accessed quickly.

We run packet scans on the NW servers (during the slow downs) to look for
users pumping large quantities of data, but have not found anything
conclusive. (The highest 'talkers' don't seem excessive).
We are fairly up-to-date with NW patches, etc. We've updated the f/w in the

Any ideas of what to try next?

Running out of ideas. Graham