FreeBSD 9.0 - kernel: mpt0: request timedout - Vmware Virtual

S

SirGollum

Hi guys,

Really hoping someone can help me. Error and config details below. I have a virtualised web server running 9.0 on zfsboot.

Specs:
8 Cores
6Gb RAM
2 x 100Gb disks (Virtual)

Every (insert random amount of days) days the server will grind to a halt and show these errors in /var/log/messages.

I initially googled the error and found - http://forums.freebsd.org/showthread.php?t=27899

Which although is a different issue, produced the same error, but the fix for that didn't work either. Does anyone have similar issues? Any ideas / advice greatly appreciated!

loader.conf

Code: 
zfs_load="YES"
vfs.root.mountfrom="zfs:ROOT"
accf_http_load="YES"
accf_data_load="YES"
hw.pci.enable_msi=0
hw.pci.enable_msix=0
# Limit queue size for ZFS
vfs.zfs.vdev.min_pending="1"
vfs.zfs.vdev.max_pending="1"

Example of errors:

Code: 
Mar  5 08:17:58 unixweb2 kernel: mpt0: request 0xffffff8000b578a0:22175 timed out for ccb 0xfffffe00040a6800 (req->ccb 0xfffffe00040a6800)
Mar  5 08:20:34 unixweb2 kernel: mpt0: attempting to abort req 0xffffff8000b578a0:22175 function 0
Mar  5 08:20:34 unixweb2 kernel: mpt0: completing timedout/aborted req 0xffffff8000b578a0:22175
Mar  5 08:20:34 unixweb2 kernel: mpt0: abort of req 0xffffff8000b578a0:0 completed
Mar  5 08:20:34 unixweb2 kernel: mpt0: request 0xffffff8000b4c140:22176 timed out for ccb 0xfffffe00b3a10000 (req->ccb 0xfffffe00b3a10000)
Mar  5 08:20:34 unixweb2 kernel: mpt0: attempting to abort req 0xffffff8000b4c140:22176 function 0
Mar  5 08:20:34 unixweb2 kernel: mpt0: completing timedout/aborted req 0xffffff8000b4c140:22176
Mar  5 08:20:34 unixweb2 kernel: mpt0: abort of req 0xffffff8000b4c140:0 completed
Mar  5 08:20:22 unixweb2 proftpd[52336]: 203.31.82.77 (124-170-120-172.dyn.iinet.net.au[124.170.120.172]) - FTP no transfer timeout, disconnected 
Mar  5 08:59:54 unixweb2 kernel: mpt0: request 0xffffff8000b48cf0:27524 timed out for ccb 0xfffffe00040a6800 (req->ccb 0xfffffe00040a6800)
Mar  5 08:59:54 unixweb2 kernel: mpt0: attempting to abort req 0xffffff8000b48cf0:27524 function 0
Mar  5 08:59:54 unixweb2 kernel: mpt0: completing timedout/aborted req 0xffffff8000b48cf0:27524
Mar  5 08:59:54 unixweb2 kernel: mpt0: abort of req 0xffffff8000b48cf0:0 completed
Mar  5 08:59:54 unixweb2 kernel: mpt0: request 0xffffff8000b55140:27525 timed out for ccb 0xfffffe00047ab000 (req->ccb 0xfffffe00047ab000)
Mar  5 08:59:54 unixweb2 kernel: mpt0: attempting to abort req 0xffffff8000b55140:27525 function 0
Mar  5 08:59:54 unixweb2 kernel: mpt0: completing timedout/aborted req 0xffffff8000b55140:27525
Mar  5 08:59:54 unixweb2 kernel: mpt0: abort of req 0xffffff8000b55140:0 completed
Mar  5 08:59:54 unixweb2 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 10879, size: 4096
Mar  5 08:59:54 unixweb2 last message repeated 10 times
Mar  5 08:59:54 unixweb2 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 428, size: 49152
Mar  5 08:59:54 unixweb2 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 10879, size: 4096
Mar  5 08:59:54 unixweb2 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 428, size: 49152
Mar  5 08:59:54 unixweb2 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 10879, size: 4096
Mar  5 08:59:54 unixweb2 kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 428, size: 49152
 
I had a read of that post and found that

Code: 
hw.pci.enable_msi=0
hw.pci.enable_msix=0

Seemed to help for that particular issue, but unfortunately hasn't helped me.
 
If you are running FreeBSD 9, I believe you are running ESXi 5.1 then.

SirGollum said:
I had a read of that post and found that

Code: 
hw.pci.enable_msi=0
hw.pci.enable_msix=0

Seemed to help for that particular issue, but unfortunately hasn't helped me.
Click to expand...
 
Then you are running unsupported configuration - 9.0 is supported officially in ESXi 5.1. The mpt0 suffers a lot under high IO in freebsd FreeBSD - even background fsck on a huge directory with many files can bring the system down :\

SirGollum said:
In this instance, it's 9.0 on Esxi 5.0.
Click to expand...
 
In my case it seems so - the virtual machine on the server is running a couple of days without issue. But if you have huge disks with many files and directories I suggest to turn off background fsck - it can cause the virtual machine to reset itself :(.

SirGollum said:
Oh wow, thank you for your reply!

Do you know if the the issue has been solved in ESXi 5.1?
Click to expand...
 
You must log in or register to reply here.
Back
Top