xfs
[Top] [All Lists]

Re: server crashing

To: Artur Makówka <juice@xxxxxxxxxxxxx>
Subject: Re: server crashing
From: David Chinner <dgc@xxxxxxx>
Date: Fri, 14 Apr 2006 13:54:22 +1000
Cc: David Chinner <dgc@xxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: <443F136F.7090201@xxxxxxxxxxxxx>
References: <443627B1.5090100@xxxxxxxxxxxxx> <20060410015916.GK2732@xxxxxxxxxxxxxxxxx> <443B60E8.6070004@xxxxxxxxxxxxx> <20060412020427.GI1484909@xxxxxxxxxxxxxxxxx> <443E2F64.8010703@xxxxxxxxxxxxx> <443F136F.7090201@xxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.4.2.1i
On Fri, Apr 14, 2006 at 05:13:51AM +0200, Artur Makówka wrote:
> Artur Makówka napisal(a):
> >David Chinner napisal(a):
> >>On Tue, Apr 11, 2006 at 09:55:20AM +0200, Artur Makówka wrote:
> >>>and why xfs_repair didnt repaired it ?
> >>
> >>xfs_repair doesn't check the free space btrees, it simply rebuilds
> >>them from scratch. Hence it won't warn about a corrupted AGF btree
> >>during repair. However, after a repair they should be consistent.
> >>
> >>OTOH, xfs_check will actually check the AGF btrees for corruption
> >>and consistency. Can you run xfs_check on the filesystem after one of
> >>these errors both before and after you run xfs_repair, and post
> >>the output?
> >
> >i can't run it so often, because to run it i think i have to unmount my 
> >filesystem. and of course this means longer downtime for my users, which 
> >for now i just can't do.

You can run xfs_check -n on a mounted filesystem, but it can give spurious
errors in that case. if you do run it on a mounted filesystem, make sure
you run 'sync' first....

> >Maybe there is some kind of bug with this situation, because i forgot to 
> >tell you one thing: i have one more XFS partition that is inside this 
> >one big XFS partition.
> 
> guess it's not. it crashed again after i removed loop-mounted XFS 
> partition - so that's not it. i also ran xfs_check on unmounted system, 
> but output was empty. it just didn't say anything, and finished checking 
> after like 10-20 minutes with no output. is it normal? i did xfs_check 

No output means there is nothing wrong with the filesystem i.e. there is no
corruption on disk. That tends to implicate an in-memory corruption of some
sort. Does you machine have ECC RAM? Can you run memtest86 to determine if
there is bad memory in your machine?

> /dev/md0 1>>logs 2>>logs. And it finished with file logs with 0 bytes. 
> Should i run it with -v ? -v generates a lot of output though.

No, -v is really only useful for debugging xfs_check problems. The
command you ran will report anything it finds wrong with the filesystem.

Cheers,

Dave.
-- 
Dave Chinner
R&D Software Enginner
SGI Australian Software Group


<Prev in Thread] Current Thread [Next in Thread>