Problems with ESX 3.5 on IBM x3650 Server - rolling I/O sector errors

[dingo]

Hi,

I have been watching this site for updates and solutions. We are having problems with rolling intermittant I/O sector errors always from sda. We have a IBM x3650 with a IBM 8k ServRaid server and 6 x 750gb SATA disks. Was orginally installed with ESX 3.0, upgraded to 3.02 now upgraded to ESX 3.5 Update 1. (up to April 10 release date).

We have upgraded the Motherboard BIOS firmware to 1.09 and the ServeRaid BIOS / Firmware to 5.2 latest level .

Still same - issue ESX runs for 1 - 14 days then starts issuing rolling I/O errors on varying sectors - always sda. We have run media verify on all the disks, CPU / Memory diags over 48hrs - all clean.

This server really can't be put into production - as it may crash at any time. Server and VM purchased from IBM. IBM not offering any support and VMware won't talk to us.(OEM support?).

The other posts report issues on same IBM hardware.

Any updates / ideas would be great!

Thanks <!-- BEGIN attachments ><! END attachments ><! BEGIN content details -->

[kjb007]

What kind of RAID is implemented on that array volume? Maybe there is a problem in the stripe routine that the LUN is created out of. Is it possible to clone the LUN on the array side and use the new one instead?

-KjB

[Rumple]

While i have not seen those errors, one of the things I always do is purchase the advanced software support from IBM. Then its IBM's problem to resolve any software issues on IBM hardware.

I hate getting into the us and them conversations.

When IBM can't resolve the issue then I make them take back the hardware and replace it with a suitable replacement model.

[dingo]

Thanks for the reply,

The drives are bound as 2 Raid5 arrays - each 1.8TB. Posibly a ESX driver issue, other posts list simular problems on Dell units with SATA disks as well.

[mike_laspina]

Hi,

What have your performed on the hardware side to isolate the fault?

Have the drives been replaced?

[huna]

Have you tried to set the SATA speed to 1.5Gbit/s on the ServeRaid controller? We had a similiar problem on a x3650 with SATA drives and Linux installed.

Regards, huna

[dingo]

Thanks, I set the rate to 1.5Gbits today on the ServeRaid 8K . we'll see how this works.( may take up to 2 weeks before we know)