Hello,
I have a strange problem on all of my openldap servers (master and slaves).
In general, requests with alias always dereferenced and scope sub on an attribute indexed are working correctly and answers come in a few milliseconds.
But sometimes , during some hours, on all my openldap servers, the same queries are taking a few seconds to be returned (between 10 to 45s). During this time all queries with alias never dereferenced are working correctly.
For the openldap master, which is only used to write entries and for slurpd replication, the load average is between 0.2 and 0.5 during the problem. For others servers that have more requests, the load average is growing up.
I don't find the reasons which cause the problem. All is coming back to normal performance after the night.
I use debian etch openldap package wich provide openldap 2.3.30. My directory has around 150000 entries. I use bdb database. I have somes tuning on DB_CONFIG that I could give, if you think that it could have an interest.
For exemple : -> During problem : - with alias deref *never* $ time ldapsearch -x -LLL -H ldap://master.my.domain -b "dc=my,dc=domain" -s sub -a never uid=jean.talus.i <ldap answer> real 0m0.007s user 0m0.004s sys 0m0.000s
- with alias deref *always* $ time ldapsearch -x -LLL -H ldap://master.my.domain -b "dc=my,dc=domain" -s sub -a always uid=jean.talus.i <ldap answer> real 0m47.225s user 0m0.004s sys 0m0.000s
$ time ldapsearch -x -LLL -H ldap://slave1.my.domain -b "dc=my,dc=domain" -s sub -a always uid=jean.talus.i Time limit exceeded (3) real 0m16.148s user 0m0.000s sys 0m0.004s
and so on on all my openldap slaves servers.
-> after the night, when it's working correctly again : $ time ldapsearch -x -LLL -H ldap://master.my.domain -b "dc=my,dc=domain" -s sub -a always uid=jean.talus.i <ldap answer> real 0m0.016s user 0m0.000s sys 0m0.004s
$ time ldapsearch -x -LLL -H ldap://slave1.my.domain -b "dc=my,dc=domain" -s sub -a always uid=jean.talus.i <ldap answer> real 0m0.020s user 0m0.000s sys 0m0.004s
I don't find anything that help me on internet. Do you know how can I correct this or what can it be the cause ? Is there some docs that could help me to tune my openldap servers to correct this ?
Regards, Julien
--On December 20, 2007 1:04:01 PM +0100 "COMBES Julien - CETE Lyon/DI/ET/PAMELA" julien.combes@i-carre.net wrote:
Hello,
I have a strange problem on all of my openldap servers (master and slaves).
I don't find anything that help me on internet. Do you know how can I correct this or what can it be the cause ? Is there some docs that could help me to tune my openldap servers to correct this ?
The fact that it is sporadic and not consistent would seem to me to indicate that there's something else going on here. Have you examined what's happening to your servers during the slow times vs the fast times? I.e., more clients doing queries? etc.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
Is your issue indexing related. Are you searching items that may or may not be indexed? Are doing doing a lot of adds that take indexing time while you are doing the searches?
Sellers
On Dec 20, 2007, at 12:48 PM, Quanah Gibson-Mount wrote:
--On December 20, 2007 1:04:01 PM +0100 "COMBES Julien - CETE Lyon/DI/ET/PAMELA" julien.combes@i-carre.net wrote:
Hello,
I have a strange problem on all of my openldap servers (master and slaves).
I don't find anything that help me on internet. Do you know how can I correct this or what can it be the cause ? Is there some docs that could help me to tune my openldap servers to correct this ?
The fact that it is sporadic and not consistent would seem to me to indicate that there's something else going on here. Have you examined what's happening to your servers during the slow times vs the fast times? I.e., more clients doing queries? etc.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc
Zimbra :: the leader in open source messaging and collaboration
______________________________________________ Chris G. Sellers | NITLE Technology 734.661.2318 | chris.sellers@nitle.org AIM: imthewherd | GTalk: cgseller@gmail.com
On Dec 20, 2007, at 12:48 PM, Quanah Gibson-Mount wrote:
The fact that it is sporadic and not consistent would seem to me to indicate that there's something else going on here.
I'm agree...
Have you examined what's happening to your servers during the slow times vs the fast
times?
I.e., more clients doing queries? etc.
All the servers are used differntly : differents applications from differents networks.
The problem happens simultaneously on all servers but load-average and cpu-usage are normal on all servers during the problem. Only servers used by applications who make a massive usage of alias deref always have there load-average increase during the problem
Le 20.12.2007 20:03, Chris G. Sellers a écrit :
Is your issue indexing related. Are you searching items that may or may not be indexed?
I don't think that the issue is indexing related because the request is done quickly when it is done with alias deref never.
Are doing doing a lot of adds that take indexing time while you are doing the searches?
I haven't see a massive add of entries during the problem.
Regars, Jlien
openldap-software@openldap.org