Skip to contentSkip to conversation

xfstests 065 failures

DISCUSSION Alex Elder 2 participants 2 messages Dec 1, 2010 – Dec 2, 2010

lore
Share & tools
lore

Alex Elder

author Reviewed-by Reviewed-by Christoph Hellwig
#0 lore raw

Dave, you were asking on IRC about test 065 failures. I asked Bill Kendall about it and he bisected to find that the commit below seems to be where the problems started. I believe the problem is that one of the times is not updated properly when renaming the file "addedfile4". Here are the commands that might affect that file in test 065:

mv addeddir4/addedfile5 addeddir4/addedfile4 mv addeddir4 addeddir6

I glanced at the commit and saw nothing obviously
wrong, but at the moment I can't really dig into
it any deeper so I thought I'd report what Bill
found so others could look.

-Alex

Quoted 1 line

From dcd79a1423f64ee0184629874805c3ac40f3a2c5 Mon Sep 17 00:00:00 2001

From: Dave Chinner <[email protected]>
Date: Tue, 28 Sep 2010 12:27:25 +1000
Subject: [PATCH] xfs: don't use vfs writeback for pure metadata modifications

Under heavy multi-way parallel create workloads, the VFS struggles to write back all the inodes that have been changed in age order. The bdi flusher thread becomes CPU bound, spending 85% of it's time in the VFS code, mostly traversing the superblock dirty inode list to separate dirty inodes old enough to flush.

We already keep an index of all metadata changes in age order - in the AIL - and continued log pressure will do age ordered writeback without any extra overhead at all. If there is no pressure on the log, the xfssyncd will periodically write back metadata in ascending disk address offset order so will be very efficient.

Hence we can stop marking VFS inodes dirty during transaction commit or when changing timestamps during transactions. This will keep the inodes in the superblock dirty list to those containing data or unlogged metadata changes.

However, the timstamp changes are slightly more complex than this - there are a couple of places that do unlogged updates of the timestamps, and the VFS need to be informed of these. Hence add a new function xfs_trans_ichgtime() for transactional changes, and leave xfs_ichgtime() for the non-transactional changes.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#1 lore raw
Quoted 8 lines · Alex Elder

Dave, you were asking on IRC about test 065 failures. I asked Bill Kendall about it and he bisected to find that the commit below seems to be where the problems started. I believe the problem is that one of the times is not updated properly when renaming the file "addedfile4". Here are the commands that might affect that file in test 065:

Quoted 7 lines

mv addeddir4/addedfile5 addeddir4/addedfile4 mv addeddir4 addeddir6

I glanced at the commit and saw nothing obviously wrong, but at the moment I can't really dig into it any deeper so I thought I'd report what Bill found so others could look.

Yeah, I know that this patch was the cause, and IIRC it only affects a hard linked directory. What I found is that the iterative dump actually contains all the correct changes (the restore part of the test does not fail) - it's just that the dump image table of contents does not contain the all the changes. I haven't tracked down why the dump TOC does not contain a modification that is actually in the dump yet....

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs