Re: freezes with reiser4 in a raid1 with 2.6.11-rc5-mm1

From: Brad Campbell
Date: Sat Mar 05 2005 - 04:30:00 EST


Florian Engelhardt wrote:

Neat trick which I only discovered in desparation last week when
battling a RAID lockup on the -rc4-mm1 kernel on a remote box.

I was also having hard lockup issues, but reseating all my PCI cards
appear to have rectified that one.


Well, there are not much PCI-Cards in this server and reseating them
didnt fix it.

Sorry, I was just pointing out what "appeared" to solve my hard-lock problems, I was not suggesting it as a cure for yours.

As root. echo b > /proc/sysrq-trigger

Of course only if you have alt-sysrq built in.


Thanks for that, i was able to reboot the machine with that trick, but
i couldnt find anything bad in the messages file.

I made some further tests with the server:
Deactivating the raid, and formating the hd´s (hdc and hdd) with
reiser4, mounting them and sharing them via nfs and ftp worked great, no
freezes, no reboots, everything perfect, even the performance.
But as soon, as i activated the raid, the server freezed, or rebooted.


A complete hard lock appears to be very rare these days with kernel bugs. It may be tickling a hardware bug somewhere. My machine was only hard-locking when I was writing to the array. A complete lock-up or reboot really does sound more hardware like. Have you tried running something like memtest86?
I found after a couple of hours of memtest86 my box would lock solid, which eliminated the linux kernel from the equation completely.

I'm running ext3 on all my machines, so I can't help with reiser at all.
I'm running a largish raid5 on 2.6.10-bk10 and a fairly large raid6 on 2.6.11-rc5-bk3. I had problems with raid6 on 2.6.11-rc4-mm1 causing raid subsystem lockups, but nothing that precluded me from rebooting with the sysrq-trigger.

Regards,
Brad
--
"Human beings, who are almost unique in having the ability
to learn from the experience of others, are also remarkable
for their apparent disinclination to do so." -- Douglas Adams
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/