AsyncIO timeout on SAN and Linux machine disk become read-only

[dst]

Hi,

I have 4 ESX 3.0.1 in a cluster attached to a DS4800 with QLA2422 (rev2) cards...

Some times a have errors in the vmkernel log on all of my ESX \!!! After that some of my linux machine crashed because disk are readonly \!!!

Logs look like:

Mar 3 02:01:31 vi3-3 vmkernel: 16:11:00:11.354 cpu2:1040)SCSI: 3731: AsyncIO timeout (5000); aborting cmd w/ sn 810116, handle 1159/0x7202898

Mar 3 02:01:31 vi3-3 vmkernel: 16:11:00:11.354 cpu2:1040)LinSCSI: 3596: Aborting cmds with world 1024, originHandle 0x7202898, originSN 810116 from vmhba1:1:0

Mar 3 02:01:31 vi3-3 vmkernel: 16:11:00:11.354 cpu2:1040)<6>qla24xx_abort_command(1): handle to abort=1118

Mar 3 02:01:31 vi3-3 vmkernel: 16:11:00:11.354 cpu2:1040)LinSCSI: 2604: Forcing host status from 2 to SCSI_HOST_OK

Mar 3 02:01:31 vi3-3 vmkernel: 16:11:00:11.354 cpu2:1040)LinSCSI: 2606: Forcing device status from SDSTAT_GOOD to SDSTAT_BUSY[/i]

This log repeating every 10s for 2mn env. and after everythings return to the normal state.

I have looked in my DS4800 logs...nothing, in my SAN switch...nothing...

Anybody have an idea ?

[Svalovs_kommun]

Hi,

Did you ever resolve this? We have the same problem.

Rutger

[christianZ]

Maybe that can help:

http://www.vmware.com/community/thread.jspa?threadID=79659&tstart=0

[dst]

Hi,

YES, i have changed le SCSI virtual card on my Linux machine to buslogic, no more ReadOnly problem \!!!

BUT my problem is:

WHAT is AsyncIO Error ! because sometimes it's my VMFS partitions who goes READONLY !!!! so ALL my vm crash !!!!!

I have well tested my SAN, upgraded to le last firmware...same problem...

Anyone have an idea ?

[dst]

After un very long time...and many test, the problem seems to be over !

IBM say to me to Set the maximum queue depth for the qlogics cards to 64

and to Set Disk.SchedNumReqOutstanding to 64 on vmware.

Problem disapears \!!!