Debian / PE2950 / Xen / SCSI Reset / Offline

[Ryan Kish]

Hello list,

I have been having an I/O issue with one of my Debian machines running Xen.

Hardware
Server: PE2950
RAID: PERC5
Config: 2 RAID5 VD
Firmware: 1.03.40-0316

Software
OS: Debian Etch w/ Xen
Kernel: 2.6.18-6-xen-686

In a nutshell, during time of moderate to high I/O, the system starts to experience SCSI timeouts before the PERC5 goes offline and the box steps out for lunch. Thus far, I have only been able to recover via a reboot. It seems there was plenty of chatter on this in 2007, I have looked many threads on this list, however this thread does a very good job of describing the issue I am experiencing: http://lists.us.dell.com/pipermail/linux-poweredge/2007-April/030577.html

As far as I can tell, the megasas driver is still under heavy development and I am aware that many of the updates are not in the kernel I currently have deployed. I have tried moving my timeouts (180) and nr (16) requests in an attempt to stabilize the box, however it does not seem to be helping.

Is anyone else having this issue? Any ideas on how I can actually get this box to the point of stability? Any advice would be greatly appreciated.

Thanks,
Ryan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.us.dell.com/pipermail/linux-poweredge/attachments/20080208/b37e4d0c/attachment.htm