Stress testing and TIMEOUT - WRITE_DMA

[Anthony Chavez]

Greetings, freebsd-questions!

I've got a number of machines to deploy in very critical locations
*very* soon, so I'd appreciate any expedient responses that I can get
about this.

I have been affected by the recent issues surrounding the recent ATA
driver changes in the 5.x branch.  I'm currently tracking RELENG_5_4
(FreeBSD 5.4-RELEASE-p6) on systems with ICH4 and ICH6 UDMA controllers.

I recently applied Soeren's patch [1] on an ICH4 system, and put a
significant write load (/usr/ports/sysutils/stress -i4 -d4) on it for
almost 2 weeks.  Here are the results:

Aug 13 21:10:01 witproto sudo:      acc : TTY=ttyp5 ; PWD=/usr/home/acc ; USER=root ; COMMAND=/usr/local/bin/stress -i 4 -d 4
Aug 19 21:31:14 witproto kernel: ad0: TIMEOUT - WRITE_DMA retrying (2 retries left) LBA=7879615
Aug 21 07:59:57 witproto kernel: ad0: TIMEOUT - WRITE_DMA retrying (2 retries left) LBA=3775775
Aug 23 23:01:41 witproto kernel: ad0: TIMEOUT - WRITE_DMA retrying (2 retries left) LBA=5560159
Aug 25 12:06:50 witproto kernel: ad0: TIMEOUT - WRITE_DMA retrying (2 retries left) LBA=5786623

FWIW, my hardware is:

atapci0: <Intel ICH4 UDMA100 controller> port 0xffa0-0xffaf,0x376,0x170-0x177,0x3f6,0x1f0-0x1f7 irq 18 at device 31.1 on pci0
ad0: 76293MB <SAMSUNG SP0802N TK100-28> at ata0-master UDMA100

During the test, the drive remained in UDMA100 mode and throughput
varied from 18ish to 50ish MB/s (eyeball average approx. 20-25 MB/s).
System load is 0.00 0.06 0.45 after killing stress.

My question is simply this: is the fact that I received 4 TIMEOUT
warnings in the space of roughly 2 weeks significant cause for concern?

[1] http://people.freebsd.org/~sos/ATA/

-- 
Anthony Chavez                                 http://anthonychavez.org/
mailto:acc at anthonychavez.org         jabber:acc at jabber.anthonychavez.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 477 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-questions/attachments/20050826/6828619b/attachment.bin