[Date Prev][Date Next] [Chronological] [Thread] [Top]

RE: performance problem with sldap

yes, this is totally the same thing I experience (though not the same output as vmstat), openldap doesn't seem to close the connections properly and I have to do a restart to force it to close the connections. Typical errors in the log (debug level 256)

daemon: accept(7) failed errno=24 (Too many open files)


jm

At 11:08 AM 11/5/2001 -0800, you wrote: 
Hmm, Interesting  I'm running 5,000 queries sequentially one after another.
There are 4,000 dynamic ports , and after 3900 searches
I'm getting an errors. When I check netstat I see a lot of following entries
:

tcp        0      0 localhost.localdom:3186 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3185 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3184 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3183 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3182 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3181 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3180 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3179 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3178 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3177 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3176 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3175 localhost.localdom:ldap
TIME_WAIT
tcp        0      0 localhost.localdom:3174 localhost.localdom:ldap
TIME_WAIT

Looks like ldapsearch not properly closing connection .

-Aleks

-----Original Message-----
From: Justin Hahn [mailto:jeh@profitlogic.com]
Sent: Monday, November 05, 2001 10:58 AM
To: 'Jan-Michael Ong'; Aleks Sheynkman
Cc: openldap-software@OpenLDAP.org
Subject: RE: performance problem with sldap


Untrue. The libdb utils have an upgrade tool. Backup your databases first of
course. On debian it's /usr/bin/db3_upgrade - don't know about other
distros.

--jeh

> -----Original Message-----
> From: Jan-Michael Ong [mailto:jmong@adobe.com]
> Sent: Monday, November 05, 2001 1:44 PM
> To: Aleks Sheynkman
> Cc: openldap-software@OpenLDAP.org
> Subject: RE: performance problem with sldap
>
>
> I don't think so... The only way I can think off is that you
> do a dump of
> your database as an LDIF (using slapcat) and importing back
> the LDIF into
> the LDAP (using slapadd)
>
> good luck
>
> jm
>
> At 10:11 AM 11/5/2001 -0800, you wrote:
> >Is there are way to migrate database from version to version?
> >
> >Thanks !
> >-Aleks
> >
> >-----Original Message-----
> >From: Jan-Michael Ong [mailto:jmong@adobe.com]
> >Sent: Monday, November 05, 2001 10:08 AM
> >To: Aleks Sheynkman
> >Cc: openldap-software@OpenLDAP.org
> >Subject: RE: performance problem with sldap
> >
> >
> >You may be running into the sockbuf_max_incoming  limit,
> adjust this in
> >your slapd.conf. (thanks Kurt Zeilenga)
> >
> >here's a sample
> >
> >   Define global ACLs to disable default read access.
> >
> ># Do not enable referrals until AFTER you have a working directory
> ># service AND an understanding of referrals.
> >#referral       ldap://root.openldap.org
> >
> >pidfile         /usr/local/var/slapd.pid
> >argsfile        /usr/local/var/slapd.args
> >loglevel        256
> >sizelimit       5
> >timelimit       3600
> >schemacheck     on
> >sockbuf_max_incoming    1234567  <--- adjust this to a
> higher number or
> >define it if you do not have it defined.
> >
> >I don't know what causes it though. Also I upgraded from
> Openldap-2.0.11 to
> >Openldap-2.0.18 and db-3.2.9 to db 3-3.11 that seemed to solve other
> >problems
> >
> >good luck
> >
> >jm
> >
> >
> >
> >At 09:40 AM 11/5/2001 -0800, you wrote:
> > >Does anyone have any ideas what might cause this problem?
> Please help..!!
> > >
> > >
> > >Thanks a lot in advance !
> > >
> > >-Aleks
> > >
> > >-----Original Message-----
> > >From: Aleks Sheynkman
> > >Sent: Monday, November 05, 2001 8:53 AM
> > >To: Aleks Sheynkman; 'Jonghyuk Choi'; 'res0k3ja@verizon.net'
> > >Cc: 'openldap-software@OpenLDAP.org'
> > >Subject: RE: performance problem with sldap
> > >
> > >
> > >I ran the same test on another server which has only 200
> entries. Basically
> > >same behavior is happening after 3,000 continues searches
> ldapsearch stat
> > >throwing "Can't contact LDAP server" errors.
> > >
> > >Thanks !
> > >
> > >-Aleks
> > >
> > >-----Original Message-----
> > >From: Aleks Sheynkman
> > >Sent: Monday, November 05, 2001 7:45 AM
> > >To: 'Jonghyuk Choi'; 'res0k3ja@verizon.net'
> > >Cc: openldap-software@OpenLDAP.org
> > >Subject: RE: performance problem with sldap
> > >
> > >
> > >I increased
> > >from :
> > >cachesize 100000000
> > >dbcachesize 320000000
> > >
> > >to
> > >cachesize 1000000000
> > >dbcachesize 1000000000
> > >sizelimit 5000000
> > >
> > >Also I added cachesize. Server seems to be up and running
> all the time.
> >When
> > >I do 5,000 searches , first 3,000 goes very fast and then it starts
> >printing
> > >error messages "ldap_bind: Can't contact LDAP server".
> > >
> > >I running server following platform:
> > >   Linux 6.2.
> > >    2x750 MHz
> > >    2GB RAM
> > >
> > >Thanks !
> > >
> > >-Aleks
> > >-----Original Message-----
> > >From: Jonghyuk Choi [mailto:jongchoi@us.ibm.com]
> > >Sent: Monday, November 05, 2001 6:28 AM
> > >To: Aleks Sheynkman
> > >Cc: openldap-software@OpenLDAP.org
> > >Subject: Re: performance problem with sldap
> > >
> > >
> > >
> > >It would be possible that slapd went down running short of memory.
> > >The entry cache size looks large for 200K directory,
> > >even though slapd allocates cache entries on demand.
> > >The entry cache is designated as the number of entries
> > >while the db cache is as the number of bytes.
> > >- Jong
> > >=======================
> > >IBM T. J. Watson Research
> > >Enterprise Linux Group, Linux Technology Center
> > >jongchoi@us.ibm.com
>