FS#17469 - [kernel26] aufs2 Union unmounting sometimes takes very long

Attached to Project: Arch Linux
Opened by Evangelos Foutras (foutrelis) - Friday, 11 December 2009, 23:35 GMT
Last edited by Ionut Biru (wonder) - Sunday, 28 February 2010, 00:34 GMT
Task Type Bug Report
Category Upstream Bugs
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Architecture All
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:

The problem is very noticeable on failed (or interrupted) builds using makechrootpkg. I posted a message [1] on the arch-general mailing list regarding this issue and Ionut Biru was able to confirm its existence.

I'm running x86_64, if that matters. I'm not sure if this issue manifests itself on i686.

Additional info:

* package version(s)
- kernel26 2.6.32-1
- aufs2 2.6.32_20091203-1.

* config and/or log files etc.

- Relevant lines from kernel.org are attached.

Steps to reproduce:

Try building a package in a chroot with makechrootpkg and then interrupting it during the build phase. It may not occur on the first time or with certain packages (it happened though when I was compiling cherokee 0.99.33-1 earlier).

----
[1] http://mailman.archlinux.org/pipermail/arch-general/2009-December/009620.html
   aufs2-kernel.log (17.4 KiB)
This task depends upon

Closed by  Ionut Biru (wonder)
Sunday, 28 February 2010, 00:34 GMT
Reason for closing:  Won't fix
Comment by Allan McRae (Allan) - Friday, 11 December 2009, 23:54 GMT
I run the x86_64 kernel and aufs2 on an i686 system and do not notice this. So it may dependent on something else?
Comment by Andrea Scarpino (BaSh) - Friday, 11 December 2009, 23:59 GMT
this happen here too.
I am on i686. partition is ext4 formatted
Comment by Allan McRae (Allan) - Saturday, 12 December 2009, 00:01 GMT
Hmmm, the reporter is using ext4 too. I use ext3. Maybe fs dependent?
Comment by Ionut Biru (wonder) - Saturday, 12 December 2009, 00:06 GMT
ext4 here too. at boost rebuilds i got sometimes 20/30 min cleanups even the packages compiles without any problem.
Comment by Nezmer (Nezmer) - Tuesday, 15 December 2009, 13:17 GMT
If all this started with 2.6.32, Maybe trying mounting with "nobarrier" will show us If the issue is caused by the new behaviour of ext4.
Comment by Ionut Biru (wonder) - Wednesday, 23 December 2009, 13:18 GMT
seems that adding nobarrier fix this "issue"
Comment by Evangelos Foutras (foutrelis) - Thursday, 24 December 2009, 11:06 GMT
I mounted the ext4 partition (where the chroots lie) with nobarrier and I confirm Ionut's experience; no slowdowns are noticed anymore.

I guess this means that the "problem" isn't in the aufs2 package? :|
Comment by Dave Reisner (falconindy) - Monday, 11 January 2010, 20:55 GMT
This happens on btrfs as well. Confirm that the 'nobarrier' option helps when using ext4.
Comment by Gerardo Exequiel Pozzi (djgera) - Sunday, 28 February 2010, 00:21 GMT
  • Field changed: Status (Assigned → Waiting on Response)
status with latest 2.6.33?
Comment by Ionut Biru (wonder) - Sunday, 28 February 2010, 00:23 GMT
we are not using aufs2 anymore in devtools and i think it can be closed

Loading...