[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

fsck journal replay times (was Re: Is this list active?)



On Wed, Jul 11, 2001 at 04:05:45PM -0500, John Hiemenz wrote:
> I subscribed this morning but have no traffic.  Sometimes no traffic is 
> a good thing.  Since I've gone to ext3 on my 2.4.5 kernel, the only 
> time I ever had problems was when the filesystems were not unmounted 
> cleanly and the journal was lost.  About 60 minutes of searching and 
> reading and I found out how to recreate the journal.

You shouldn't need to do this.  After a crash with ext2, e2fsck will
repair the fs so that ext3 can mount it again (in journaling mode.)

On an almost (but not entirely) unrelated tangent, has anyone else
running ext3 over RAID-1 seen really bad journal replay times?  e2fsck
prints "replaying journal" and then grinds away at the disk for a good
thirty to forty-five seconds.  My (uneducated!) guess is that fsck is
competing for disk bandwidth with background mirror reconstruction.

Disks are UDMA-33 running 2.4.5-10 from Rawhide.  /usr is a pair of
10gb partitions on separate channels running in RAID-1.  ext3 on any
non-RAID partition seems to recover in no time at all.

Any thoughts?  Is this a bug?  XFS didn't appear to suffer from this
problem, but AFAIK xfs only does kernel-side journal replay.

-ed





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]