public inbox for glibc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug nscd/292] nscd performance drop with case insensitive attributes
[not found] <20040729072046.292.muehlenhoff@univention.de>
@ 2004-08-04 8:18 ` jakub at redhat dot com
2004-08-17 8:55 ` muehlenhoff at univention dot de
` (4 subsequent siblings)
5 siblings, 0 replies; 6+ messages in thread
From: jakub at redhat dot com @ 2004-08-04 8:18 UTC (permalink / raw)
To: glibc-bugs
------- Additional Comments From jakub at redhat dot com 2004-08-04 08:17 -------
This should be fixed in glibc CVS:
2004-07-27 Ulrich Drepper <drepper@redhat.com>
* nscd/grpcache.c (cache_addgr): If necessary, add entry also
under the name the user provided.
* nscd/pwdcache.c (cache_addpw): Likewise.
--
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution| |FIXED
http://sources.redhat.com/bugzilla/show_bug.cgi?id=292
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug nscd/292] nscd performance drop with case insensitive attributes
[not found] <20040729072046.292.muehlenhoff@univention.de>
2004-08-04 8:18 ` [Bug nscd/292] nscd performance drop with case insensitive attributes jakub at redhat dot com
@ 2004-08-17 8:55 ` muehlenhoff at univention dot de
2004-08-18 20:01 ` drepper at redhat dot com
` (3 subsequent siblings)
5 siblings, 0 replies; 6+ messages in thread
From: muehlenhoff at univention dot de @ 2004-08-17 8:55 UTC (permalink / raw)
To: glibc-bugs
------- Additional Comments From muehlenhoff at univention dot de 2004-08-17 08:55 -------
Unfortunately this is not fixed with the changes. The patches slow down the process
tremendously, but show no principal difference. I'm therefore reopening the bug.
See attached results of the same system with 2,800 users in the group "Domain Users":
Again the results with the disabled nscd:
testing:~# time getent group "Domain Users"
foo1, ... , foo2736 (output snipped for all examples)
real 0m1.964s
user 0m1.470s
sys 0m0.020s
testing:~#
testing:~# time getent group "Domain users"
real 0m1.965s
user 0m1.440s
sys 0m0.050s
testing:~#
And now the results with nscd enabled:
testing:~# time getent group "Domain Users"
real 0m4.969s
user 0m0.010s
sys 0m0.000s
testing:~#
testing:~# time getent group "Domain Users"
real 0m0.006s
user 0m0.000s
sys 0m0.010s
testing:~#
So, likewise as without the patches nscd works fine for "Domain Users". Now the
results for the request for "Domain users" (also with the enabled nscd):
testing:~# time getent group "Domain Users"
real 1m17.391s
user 0m0.000s
sys 0m0.000s
testing:~#
testing:~# time getent group "Domain Users"
real 1m16.276s
user 0m0.000s
sys 0m0.020s
So, as without the patches nscd does not speed up subsequent cache hits but with the
patches it is even 7 times slower.
--
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |REOPENED
Resolution|FIXED |
http://sources.redhat.com/bugzilla/show_bug.cgi?id=292
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug nscd/292] nscd performance drop with case insensitive attributes
[not found] <20040729072046.292.muehlenhoff@univention.de>
2004-08-04 8:18 ` [Bug nscd/292] nscd performance drop with case insensitive attributes jakub at redhat dot com
2004-08-17 8:55 ` muehlenhoff at univention dot de
@ 2004-08-18 20:01 ` drepper at redhat dot com
2004-09-26 21:33 ` drepper at redhat dot com
` (2 subsequent siblings)
5 siblings, 0 replies; 6+ messages in thread
From: drepper at redhat dot com @ 2004-08-18 20:01 UTC (permalink / raw)
To: glibc-bugs
------- Additional Comments From drepper at redhat dot com 2004-08-18 20:01 -------
Unless you provide an easy to use receipe to reproduce the problems nothing will
happen. I.e., send a script which creates the infrastructure and then describe
what you do.
--
What |Removed |Added
----------------------------------------------------------------------------
Status|REOPENED |WAITING
http://sources.redhat.com/bugzilla/show_bug.cgi?id=292
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug nscd/292] nscd performance drop with case insensitive attributes
[not found] <20040729072046.292.muehlenhoff@univention.de>
` (2 preceding siblings ...)
2004-08-18 20:01 ` drepper at redhat dot com
@ 2004-09-26 21:33 ` drepper at redhat dot com
2004-09-27 12:02 ` muehlenhoff at univention dot de
2004-10-07 11:30 ` drepper at redhat dot com
5 siblings, 0 replies; 6+ messages in thread
From: drepper at redhat dot com @ 2004-09-26 21:33 UTC (permalink / raw)
To: glibc-bugs
------- Additional Comments From drepper at redhat dot com 2004-09-26 21:33 -------
Well? No reply? Is there no problem after all?
--
http://sources.redhat.com/bugzilla/show_bug.cgi?id=292
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug nscd/292] nscd performance drop with case insensitive attributes
[not found] <20040729072046.292.muehlenhoff@univention.de>
` (3 preceding siblings ...)
2004-09-26 21:33 ` drepper at redhat dot com
@ 2004-09-27 12:02 ` muehlenhoff at univention dot de
2004-10-07 11:30 ` drepper at redhat dot com
5 siblings, 0 replies; 6+ messages in thread
From: muehlenhoff at univention dot de @ 2004-09-27 12:02 UTC (permalink / raw)
To: glibc-bugs
------- Additional Comments From muehlenhoff at univention dot de 2004-09-27 12:02 -------
Sorry for the late reply. We provide an easy test case for you: The problem can only be
properly reproduced with LDAP, as access through "compat" is way too fast for reliable
performance measures. There are two easy ways for you to check the problem:
1. Two shell scripts, that generate the necessary LDIF files for 2000 users, one for the
user and the group data:
http://bitz150.bitz.briteline.de/nscd-glibc-bug-292-generate-group-ldif.sh
http://bitz150.bitz.briteline.de/nscd-glibc-bug-292-generate-user-ldif.sh
Add the LDIFs with the following command where LDAPPASSWD is your LDAP admin
password:
ldapadd -x -c -D cn=admin,dc=nodomain -w LDAPPASSWD -f group.ldif
ldapadd -x -c -D cn=admin,dc=nodomain -w LDAPPASSWD -f user.ldif
2. Another, even easier way, is to just import the prepared slapcat output of the two
scripts above:
http://bitz150.bitz.briteline.de/nscd-glibc-bug-292-complete-slapcat-data.bz2
If you have further requests, just follow up and we'll provide it ASAP.
--
http://sources.redhat.com/bugzilla/show_bug.cgi?id=292
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug nscd/292] nscd performance drop with case insensitive attributes
[not found] <20040729072046.292.muehlenhoff@univention.de>
` (4 preceding siblings ...)
2004-09-27 12:02 ` muehlenhoff at univention dot de
@ 2004-10-07 11:30 ` drepper at redhat dot com
5 siblings, 0 replies; 6+ messages in thread
From: drepper at redhat dot com @ 2004-10-07 11:30 UTC (permalink / raw)
To: glibc-bugs
------- Additional Comments From drepper at redhat dot com 2004-10-07 11:30 -------
I cannot reproduce anything like this with the current code base. I have
nothing older around and even if, it's meaningless.
Without nscd running I get these times:
$ time getent group "Domain Users" > /dev/null
real 0m0.122s
user 0m0.032s
sys 0m0.009s
$ time getent group "Domain users" > /dev/null
real 0m0.123s
user 0m0.036s
sys 0m0.017s
When I start nscd (the persistent cache removed) I get for the first and second
run these numbers:
$ time getent group "Domain Users" > /dev/null
real 0m0.132s
user 0m0.001s
sys 0m0.003s
$ time getent group "Domain Users" > /dev/null
real 0m0.003s
user 0m0.001s
sys 0m0.002s
$ time getent group "Domain users" > /dev/null
real 0m0.132s
user 0m0.000s
sys 0m0.003s
$ time getent group "Domain users" > /dev/null
real 0m0.003s
user 0m0.001s
sys 0m0.003s
The first query of "Domain users" of course had to get the data from the server
again since the "Domain Users" lookup does not have this key and nscd does not
know (and must not know) about the case insensitivity of LDAP.
I.e., if you really have problems, use a system using the current code base.
E.g., FC3t2.
--
What |Removed |Added
----------------------------------------------------------------------------
Status|WAITING |RESOLVED
Resolution| |WORKSFORME
http://sources.redhat.com/bugzilla/show_bug.cgi?id=292
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-10-07 11:30 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <20040729072046.292.muehlenhoff@univention.de>
2004-08-04 8:18 ` [Bug nscd/292] nscd performance drop with case insensitive attributes jakub at redhat dot com
2004-08-17 8:55 ` muehlenhoff at univention dot de
2004-08-18 20:01 ` drepper at redhat dot com
2004-09-26 21:33 ` drepper at redhat dot com
2004-09-27 12:02 ` muehlenhoff at univention dot de
2004-10-07 11:30 ` drepper at redhat dot com
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).