server hanging

"Castelhano, Dan" <[email protected]> Wed, 9 Apr 2003 08:47:02 -0400
Newsgroups gmane.linux.redhat.release.seawolf
Message-ID <[email protected]>
Hello,

I have a server which is hanging after about 2-4 hrs of heavy Disk i/o and
cpu utilization and running a re-compiled kernel. It's running redhat 7.1
with the 2.4.20 kernel. 

Hardware is:
Hp/Compaq DL380g3
4.6GB ram
raid5 partitions
2.8ghz xeon's

what's really odd is this only happens when the server has more then 4GB ram
AND is using the embedded 5i raid controller card found on this system. If I
run the harddrives from a PCI based Compaq raid card and 4.6gb ram, server
runs fine. If I bring the server down to 4gb ram (4x1gb chips) and use the
embedded raid card server is also fine.

But it's that combination of over 4gb RAM and the embedded raid card which
is causing problems.

Both the embedded raid card (5i plus) and the PCI card (smartarray 5302) use
the same driver (cciss) and have the same amount of cache (64mb) - although
the 5302 pci card is doing read/write cache, where as the 5i is read-only.

As a test, I setup a test server with same hardware but running 7.3 with its
original kernel - 2.4.18. With this setup, server runs fine. If I rebuild
the 2.4.20 or 2.4.19 kernels, server hangs once again. I have many systems
which are running a recompiled kernel, and have never encountered a problem
like this.

Has anyone ever encountered a similar problem with a kernel they've
compiled? I've gone through all the options in make config and nothing
seemed obvious to me which would be causing this.

I've attached my kernel 2.4.20 ".config" file in the hopes that it'll aid
someone in figuring out what could be happening.

Thank you,
-dan

 <<config>>
config (application/octet-stream, 16.3 KB) - not displayed