Alpha Linux ECC Errors

[Michael Huntingdon <hunting ix netcom com>]

I have a cluster of 12 DS10's with either 1GB or 512MB of Dataram memory.
All of the nodes appear to be less than stable in terms of the number of
errors logged. During testing some have crashed. Case heating looks to be
well within tolerance.

So far we've removed the Dataram memory and replaced it with the less dense
Compaq memory left over from the installation. Systems have gone from 512MB
of Dataram to 128MB of Compaq memory. These systems are stable showing no
ecc errors.

As of today we have replaced the Dataram memory in 7 systems which improves
the number of errors but does not eliminate them. We have also applied the
latest SRM version 5.5-7.

Anyone seen issues like this with the DS10's and Dataram or any other vendor?

regards
michael