[opensuse] Default install creates /etc/.pwd.lock in 11.1
Hello, I have noticed huge delays in many operations (e.g. "ls -l"). After some investigation it turned out that the delays are caused by nscd when user/group name lookup is done. Even worse: it is not possible to change passwords or modify user accounts. The problem is caused because the installation procedure creates a file named /etc/.pwd.lock dated from 2008-12-09 22:41. When installing with "Installation from Images" activated, this file is created by the installation of the "KDE[base]" Image. Then I tried to install with "Installation from images" deactivated, but the file was created anyway (with a different timestamp this time), thus I assume some package is broken. Unfortunately, I could not figure out which package was the cause. A simple "rm /etc/.pwd.lock" cured the problem for me. But I guess this problem deserves more investigation. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Tue, Jan 06, 2009 at 11:46:59AM +0100, Josef Wolf wrote:
I have noticed huge delays in many operations (e.g. "ls -l"). After some investigation it turned out that the delays are caused by nscd when user/group name lookup is done. Even worse: it is not possible to change passwords or modify user accounts.
What's up? nobody cares? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Wednesday, 2009-01-07 at 23:41 +0100, Josef Wolf wrote:
On Tue, Jan 06, 2009 at 11:46:59AM +0100, Josef Wolf wrote:
I have noticed huge delays in many operations (e.g. "ls -l"). After some investigation it turned out that the delays are caused by nscd when user/group name lookup is done. Even worse: it is not possible to change passwords or modify user accounts.
What's up? nobody cares?
I have that file, but I haven't noticed delays the few times i have been able to run 11.1 - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkllNosACgkQtTMYHG2NR9UIZQCePpL3N1SOCzymedX01NKoh4+r wcYAn0Cf7gnwIlOO8Szsy2Gtc3OE+drj =FM0G -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thursday 08 January 2009, Josef Wolf wrote:
On Tue, Jan 06, 2009 at 11:46:59AM +0100, Josef Wolf wrote:
I have noticed huge delays in many operations (e.g. "ls -l"). After some investigation it turned out that the delays are caused by nscd when user/group name lookup is done. Even worse: it is not possible to change passwords or modify user accounts.
What's up? nobody cares?
I have /etc/.pwd.lock here dating back to 2006, so I don't think it's related to the problem. On the other hand, nscd is very buggy and crashes at least once a day here. This looks like a problem with name resolution to me, check /etc/hosts for the localhost entry. In order to investigate this further, try "strace ls" in an empty directory and check where it spends most time. Regards -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
* auxsvr@gmail.com
I have /etc/.pwd.lock here dating back to 2006, so I don't think it's related to the problem.
+1 -- Patrick Shanahan Plainfield, Indiana, USA HOG # US1244711 http://wahoo.no-ip.org Photo Album: http://wahoo.no-ip.org/gallery2 Registered Linux User #207535 @ http://counter.li.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thu, Jan 08, 2009 at 01:17:26AM +0200, auxsvr@gmail.com wrote:
On Thursday 08 January 2009, Josef Wolf wrote:
On Tue, Jan 06, 2009 at 11:46:59AM +0100, Josef Wolf wrote:
I have noticed huge delays in many operations (e.g. "ls -l"). After some investigation it turned out that the delays are caused by nscd when user/group name lookup is done. Even worse: it is not possible to change passwords or modify user accounts.
What's up? nobody cares?
I have /etc/.pwd.lock here dating back to 2006, so I don't think it's related to the problem.
When I removed the file, the problem disappeared.
On the other hand, nscd is very buggy and crashes at least once a day here.
Hmm, I have no clue why nscd is installed at all... I have not installed it intentionally. I only learned about its existance while I was tracking down the problem.
This looks like a problem with name resolution to me, check /etc/hosts for the localhost entry.
$ grep 127 /etc/hosts 127.0.0.1 localhost 127.0.0.2 raven.my.domain raven $
In order to investigate this further, try "strace ls" in an empty directory and check where it spends most time.
Before I wrote my first mail of this thread, I have actually done strace groupadd foo and it _was_ actually hanging trying to flock(2) /etc/.pwd.lock. The problem was perfectly reproducible: I have reinstalled three times just to figure out where the file came from and to make sure that it was not me causing the problem. Now, I have re-created the file with "touch /etc/.pwd.lock", and it does _not_ hang anymore. Guess, I have to reinstall to reproduce the problem again. Anyway, a stale lock file should _not_ exist in a clean install. And even _if_ it exists, this situation should be detected and removed, IMHO. Thanks for your comment! -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thursday 08 January 2009, Josef Wolf wrote:
On Thu, Jan 08, 2009 at 01:17:26AM +0200, auxsvr@gmail.com wrote:
I have /etc/.pwd.lock here dating back to 2006, so I don't think it's related to the problem.
When I removed the file, the problem disappeared.
Weird, I still have the file here and groupadd foo returns immediately.
On the other hand, nscd is very buggy and crashes at least once a day here.
Hmm, I have no clue why nscd is installed at all... I have not installed it intentionally. I only learned about its existance while I was tracking down the problem.
It's meant to be a name service caching daemon, when it doesn't crash that is...
This looks like a problem with name resolution to me, check /etc/hosts for the localhost entry.
$ grep 127 /etc/hosts 127.0.0.1 localhost 127.0.0.2 raven.my.domain raven $
OK
In order to investigate this further, try "strace ls" in an empty directory and check where it spends most time.
Before I wrote my first mail of this thread, I have actually done
strace groupadd foo
and it _was_ actually hanging trying to flock(2) /etc/.pwd.lock. The problem was perfectly reproducible: I have reinstalled three times just to figure out where the file came from and to make sure that it was not me causing the problem.
groupadd foo simply opens /etc/.pwd.lock here, no flock occurs.
Now, I have re-created the file with "touch /etc/.pwd.lock", and it does _not_ hang anymore.
Guess, I have to reinstall to reproduce the problem again.
Anyway, a stale lock file should _not_ exist in a clean install. And even _if_ it exists, this situation should be detected and removed, IMHO.
Thanks for your comment!
If I remove /etc/.pwd.lock, groupadd creates it again. I still don't think the problem is /etc/.pwd.lock, maybe /etc/ is accessed via NFS? Regards -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (4)
-
auxsvr＠gmail.com -
Carlos E. R. -
Josef Wolf -
Patrick Shanahan