Bug 803976 - rpm and updatedb cause (temporary) system lockup
rpm and updatedb cause (temporary) system lockup
Product: Fedora
Classification: Fedora
Component: kernel (Show other bugs)
i386 Linux
unspecified Severity high
: ---
: ---
Assigned To: Kernel Maintainer List
Fedora Extras Quality Assurance
Depends On:
  Show dependency treegraph
Reported: 2012-03-16 03:58 EDT by Dov Grobgeld
Modified: 2012-07-11 13:51 EDT (History)
9 users (show)

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2012-07-11 13:51:29 EDT
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)

  None (edit)
Description Dov Grobgeld 2012-03-16 03:58:19 EDT
Description of problem:

rpm causes temporary system lockups when run. These lockups cause some X11 programs to be unresponsive (no display refresh) for time periods ~30s and then afterwards are responsive again. (Among these clients are fedora and emacs. xterm strangely enough is still responsive.) I tried to rebuild the rpm database with `rpmdb --rebuild` but it didn't help. During the rebuild rpmdb also caused these lockups. The system locks up for about 30s and is then released and this happens again after another minute or so. 

In addition, even after `rpmdb --rebuild` the rpm is unbearably slow. E.g.

[root@grower dov]# time rpm -q -a > /dev/null

real    1m13.801s
user    0m7.181s
sys     0m0.239s

`top` shows that it is not a CPU issue. `lsof` doesn't reveal anything suspicious (to me). Doing strace on rpm shows that it never gets stuck on any timeout.

A related problem is that updatedb cause a similar lockup.

Version-Release number of selected component (if applicable):

# uname -a
Linux grower #1 SMP Mon Jan 23 15:44:18 UTC 2012 i686 i686 i386 GNU/Linux
# cat /etc/redhat-release 
Fedora release 15 (Lovelock)
# rpm --version
RPM version
# updatedb --version
updatedb (mlocate) 0.24
Copyright (C) 2007 Red Hat, Inc. All rights reserved.
This software is distributed under the GPL v.2.

This program is provided with NO WARRANTY, to the extent permitted by law.

How reproducible:

Every time.

Steps to Reproduce:
1. Open firefox
2. run `rpm -q -a > /dev/null`
3. or run `updatedb`
Actual results:

Temporary system freeze.

Expected results:

Background activity without any visible impact on desktop.

Additional info:
Comment 1 Panu Matilainen 2012-03-16 04:19:43 EDT
System freezes on rpm, updatedb or such are kernel land, reassigning. See the timings at https://bugzilla.redhat.com/show_bug.cgi?id=752897#c28 with different kernels - the test-case timing goes from 2m 39s to 7m 53s (ie nearly triples) depending on which kernel is running. That's quite a regression, apparently related to heavy IO.

As for the rpm part, make sure you're using this update which migitates the issue somewhat: https://admin.fedoraproject.org/updates/FEDORA-2012-1510/rpm-
Comment 2 Dov Grobgeld 2012-03-16 04:57:11 EDT
I just installed that update and it made a *huge* difference in response time for rpm! :-)

Now we just need a similar fix for mlocate/updatedb.

But we can consider this bug closed or a duplicate.
Comment 3 Panu Matilainen 2012-03-16 05:13:29 EDT
The rpm update only helps to avoid triggering the worst-case scenario (whatever it is) by using a smaller cache for the rpm database.

The case is not really closed until somebody figures out what has caused the severe IO performance degration in the more recent kernels (again, see the figures in bug 752897).
Comment 4 Justin M. Forbes 2012-03-16 14:51:07 EDT
Is this still happening with kernel- (in updates-testing)? is actually quite old, several things have been fixed in that time, including several security issues.
Comment 5 Dov Grobgeld 2012-03-17 17:50:01 EDT
I updated to kernel- and the result was as follows:

* Running `updatedb` at the default nice level still caused firefox to be inaccessible for periods of roughly 5s. This is still much better than the behaviour I experienced before.
* Running `updatedb` at nice level 19 (nicest level) causes no detectable interference in desktop use.

Since I updated rpm I can no longer test its desktop influence. Do you need me to downgrade rpm to test without the latest workaround?
Comment 6 Josh Boyer 2012-06-07 13:57:54 EDT
Are you still seeing the updatedb issue with 2.6.43/3.3?
Comment 7 Josh Boyer 2012-07-11 13:51:29 EDT
Fedora 15 has reached it's end of life as of June 26, 2012.  As a result, we will not be fixing any remaining bugs found in Fedora 15.

In the event that you have upgraded to a newer release and the bug you reported is still present, please reopen the bug and set the version field to the newest release you have encountered the issue with.  Before doing so, please ensure you are testing the latest kernel update in that release and attach any new and relevant information you may have gathered.

Thank you for taking the time to file a report.  We hope newer versions of Fedora suit your needs.

Note You need to log in before you can comment on or make changes to this bug.