Hi, I am running a news server as well (I started this thread). Here's
a quote from an e-mail a friend sent me, who also runs a news server:
> Drop Linux like a hot rock!! It is fine for user boxes but is totally
> unsuitable for a server. Ext2fs is so unbelievably unstable I can't
> imagine why it took so long to see it.
> I started out with Linux for our news server. About 5 times I came in to
> find the machine hung and a reboot showed the file systems to be trashed.
> I at first blamed the crappyu Micropolis 9GB disks I was using. But the
> problem persisted after I dumped them in favor of Fuji's. So I did some
> more testing. If I didn't reboot every 24 hours to flush the filesystem
> and force frequent fsck's the file system crashed. If I let it go for a
> week, even if it didn't crash, a reboot would find a filesystem that even
> fsck couldn't fix. I have no doubt that the problem lies with Ext2fs.
Now, I'm not planning on dropping Linux like a hot rock -- I like it too
much -- but it is clear there is a serious problem *somewhere* that
causes file corruption.
I applied the patch tytso mailed here a few days ago, which he used to
combat AFS corruption, but to no avail. I plan to do some hardcore
debugging this weekend myself to see if I can (a) reproduce the problem
regularly, and (b) locate and possibly fix the problem.
I've had this problem with 2.0.25 and 2.0.29. I'm not sure whether
pre-patch-2.0.30-2 solves the problem or not, but I did notice a couple
inode operations now guarded by semaphores in the patch. I'll be doing
my stress testing and debugging off of the pre-patch first, and then off
of 2.0.29.
As always, any help or suggestions would be greatly appreciated.
Jeff