Research time is important (3)

[Benoit.Le_Bras@alcatel.fr]

 Hi All,

 I am trying to do a search on my database that contains about 400 000 entries

 When i do a search like

 ldapsearch ... "(&(|(VoiceMailboxId=10)(VoiceMailboxId=11)(VoiceMailboxId=12))(VPUNumber=2)(vpuSiteName=clust50))"

 the result research take 4 minutes (timelimit) in one case :

 - if the vpuSiteName exist, VPUNumber don't exist and whatever the existence of the VoiceMailboxId

 For the other case the result research take 1 sec

 My configuration in slapd.conf :

 index cn,uid,telephoneNumber eq
 index telephoneNumber2,telephoneNumber1,telephoneNumber3,telephoneNumber4,adrIMAP4 eq
 index vpuSiteName,VoiceMailboxId,VPUNumber eq
 index objectClass eq

I have watch the log file and ,when the research take the timelimit, slapd compare one by one the entries to the filter

I have regenerate the index and there is the same problem.

I use a BerkeleyDB database and openldap 2.0.4

Any information would be greatly appreciated.

Thanks in advance

Benoit