severe FS corruption with 2.4.6-pre2 + IBM jfs 0.3.4 patch
It's probably a JFS issue, but I thought I'd report this in case someone is collecting and correlating filesystem corruption messages (Alan?). Here is my sad story.
I have an Athlon 700MHZ, 256mb ram, AIC7XXX w 2/U2W drives system. I've been running JFS on a small partition for a few weeks, JFS 0.3.1 under 2.4.5, 0.3.3 with 2.4.6-pre1 and yesterday I tried 2.4.6-pre2 with jfs 0.3.4.
I rebooted into the new kernel, and noticed one of those Red Hat [FAIL] messages that pops up during the rc/init stages. It's a config problem I've had for months, nothing new, but I decided I'd take care of it. I logged into the console, ran emacs and it segfaulted. I ran 'dmesg' but there was nothing (I was expecting to see an oops). I ran emacs again, segfault. I tried logging in to another console, hang.
At this point I hit sysrq-T, and of everything that spewed by, I noticed one process (which had scrolled to far off the screen to see the PID) with a whole lot of
<dbdbdbdb> <dbdbdbdb> <dbdbdbdb> in the stack trace.
At this point I did sysrq-S-U-B and rebooted. End of story.
My filesystems were severely f***ed at this point. A total of about 60 seconds running this kernel and I was dead. My OS partition was trashed, about 1/2 of it was unrecoverable. My home partition fared better (good luck) but still had plenty of trashed inodes.
I've never had problems with any kernels before this...
That's my story,
David
Signature · 3 lines
David Mansfield (718) 963-2020 [email protected] Ultramaster Group, LLC www.ultramaster.com
Quoted 3 lines
It's probably a JFS issue, but I thought I'd report this in case someone is collecting and correlating filesystem corruption messages (Alan?). Here is my sad story.
I get as far as 'using jfs' and delete them
Quoted 6 lines · Alan Cox
Quoted 4 lines
It's probably a JFS issue, but I thought I'd report this in case someone is collecting and correlating filesystem corruption messages (Alan?). Here is my sad story.
I get as far as 'using jfs' and delete them
Understandable but FWIW they have apparently passed a night of stress-kernel (cerberus) testing on the latest jfs..
Signature · 3 lines
Jeff Garzik | Andre the Giant has a posse. Building 1024 | MandrakeSoft |
Quoted 10 lines · Jeff Garzik
Quoted 6 lines · Alan Cox
Quoted 4 lines
It's probably a JFS issue, but I thought I'd report this in case someone is collecting and correlating filesystem corruption messages (Alan?). Here is my sad story.
I get as far as 'using jfs' and delete them
Understandable but FWIW they have apparently passed a night of stress-kernel (cerberus) testing on the latest jfs..
rm -rf not working correctly is a kind of show stopper bug ATM though. Hopefully it can be fixed soon.
-Andi
Quoted 2 lines · Andi Kleen
rm -rf not working correctly is a kind of show stopper bug ATM though. Hopefully it can be fixed soon.
with too many files inside of a directory this was a bug also under AIX 3.2.5, and could be used for any kind of DoS.
Just with AIX 4.1.X this bug disappeared and we also got the possibility to use a block size smaller than 1024
Luigi