Linux VMs on ESXi 4 gives I/O errors

[AirtjE]

Hi all.

I recently bought a new server and put ESXi 4 on it. All hardware is new. It runs on a Supermicro X8STi with 12 GB DDR3 RAM and a Core i7 920 processor. I managed to install ESXi 4 easily. I came to the conclusion that it does not support RAID configurations using the onboard SATA controller, thus use a non-RAID configuration. My Western Digital WD10EVDS disks are running on native SATA AHCI mode now.

ESXi 4 installation went all well. It recognizes my ICH10 storage adapter and its disks and I was able to create VMs and run the installations.

A test installation of a simple Windows Server or workstation went good. It still works just fine.

However, a Debian VM ran for a week and then suddenly my MySQL databases became corrupted and lots of I/O erros came up. I did a fsck and found massive amounts of "Clearing orphaned inode" errors. Some of the files cannot be fixed and resulted in some services not able to start.

As a test, I installed several VMs with different Linux distros (Debian Lenny, CentOS, Ubuntu) with both x86 as x64 versions. All installations were succesful, however after first reboot and a fsck, they all returned "Clearing orphaned inode" errors. Even when I install the VMTools afterwards, I still get those errors after or even without reboots.

I checked events logged in ESXi but did not find any errors.

I checked the vmware.log and messages under SSH of the ESXi machine, found no particular problems.

I still have those Windows Server 2008, Windows 7 and a Windows XP VM running on that same machine on ESXi4. They all run fine. Even when I do a checkdisk under Windows.

The current ESXi installation is fully up-to-date (U2). I'm running out of ideas.... Any help is greatly appreciated... anyone?

[DSTAVERT]

I would run any diagnostics tools available from the manufacturer on the bare server. It is hard to do testing on the drives without loosing existing data but I would do it just to eliminate those components as problems. Motherboard firmware up to date? I would have a look through http://vm-help.com for any reference to your hardware since I will guess it isn't on the hardware compatibility list.

The logs for the individual virtual machines would be the place to look for error conditions.

[AirtjE]

Thanks for your input.

Read several articles now on vm-help etc. Although the exact version of this mainboard is not listed on the HCL or Whitebox HCL, many Supermicro mainboards simular to this one are listed and the Intel ICH10 AHCI, network cards and CPU are listed on the HCL. I expect it should work one way or another.

Checked the logfiles but don't really know what to look out for, but didn't find anything special.

I am planning to do the following: (somewhere in the upcoming 2-3 weeks hopefully, very busy period):

- BIOS upgrade to v1.0c

- BIOS reset

- BIOS try changing virtualisation settings

- BIOS try changing AHCI to IDE mode

- Try to connect it to another SATA controller on-board

Will post here if I found a solution, but meanwhile new ideas / tips are of course more than welcome

Thanks!