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

Re: Software RAID & ext3 v0.0.6b


On Mon, Mar 12, 2001 at 03:42:13PM +0000, Alan Hourihane wrote:
> I've just set up a brand new system with software raid1 (in degraded mode)
> with one IDE 20GB drive, using kernel 2.2.19pre16 with ext3 0.0.6b.

Umm, don't do that.  2.2's soft raid is incompatible with journaling
of any form, and ext3 is no exception: this _will_ break.

> But. When I plug the second disk in a start a reconstruction, sooner or later
> within seconds I get an 'ASSERT' failure at line 411 in journal.c. !
> It's this line....
> 	J_ASSERT(bh->b_jlist == 0 || bh->b_jlist == BJ_LogCtl || bh->b_jlist == BJ_IO || bh->b_jlist == BJ_Data);

This is expected.  The raid reconstruction code writes stuff from the
buffer cache to disk without permission, violating the journaling
code's write ordering and triggering the ext3 write order debugging

Raid0/linear and LVM should be OK, but raid1 and raid5 cannot be used
with journaling (neither ext3 nor reiserfs) on 2.2.

2.4's raid layer fixes this by doing the raid reconstruction outside
of the buffer cache.


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