I would like to explain again my problem that I can reproduice very
I have 3 running scripts: each script executes a specific ldapsearch and sleep 1s ...etc
each script scans all the database!
The database contains 10000 subscribers (that is 30000 dn)
I launch too a script which simulates a high CPU use!
and approximately 1h30 afterwards the ldapsearch requests are altered that is instead of 11ms per request we have 15 seconds per search request
and even if I stop my script which simulates a high CPU use (so the
CPU use becomes again normal),
my search requests take always long time (15 seconds in average)
Then if I execute the "db_recover -v" command in my database and I restart
my openldap server
Again my search requests take a correct time (11ms in average)
I don't understand what the traces of the "db_recover" command
means ?? => cf. traces_recover file in attachment
I used too the command "db_printlog" to make printable the log files but it's the same pb
So my question is: which issue/exit do I have ?
1. go on anlysing traces but how doing ?
2. try to configure BDB backend but how doing ?
3. replace my BDB backend by LDBM backend ? where can I find a ldbm backend ...?
Thanks a lot
Please read the FAQ-o-Matic before attempting to diagnose the problem. http://www.openldap.org/faq/data/cache/893.html
/opt/BerkeleyDB/bin/db_recover -v db_recover: Finding last valid log LSN: file: 12 offset 1679343 db_recover: Recovery starting from  db_recover: Recovery complete at Wed Oct 22 15:26:50 2003 db_recover: Maximum transaction ID 80000008 Recovery checkpoint  db_recover: Recovery complete at Wed Oct 22 15:26:50 2003 db_recover: Maximum transaction id 80000000 Recovery checkpoint 
#!/bin/sh nb=100 while [ $nb -gt 0 ] do while read msisdn do time /opt/openldap-clients-servers/bin/ldapsearch -x -D \ "cn=admin,ou=CONTENT,o=alcatel,c=fr" -w secret -b \ "ou=REPOSITORY,ou=CONTENT,o=alcatel,c=fr" -h \ 220.127.116.11:389 "email@example.com" homeDirectory mailQuota mailpassword maillogin cn sleep 1 done < ./pnnumber.txt done