directory-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From <Carlo.Acco...@ibs-ag.com>
Subject RE: Plea for help with search performance
Date Thu, 12 Apr 2012 12:28:13 GMT
Other things I've tried. 
Used the attribute OID instead of its literal in ads-indexattributeid

Tried copying a newly created .db file and renaming it 2.16.840.1.113730.3.1.241.db  for displayName
and adding the 
corresponding 2.16.840.1.113730.3.1.241-displayName.txt file with Its schema definition. When
the server starts up, it just deletes the files. 

Also I tried building from the trunk to test that code base but can't get past goal apacheds-interceptors-admin.

Any work around you might provide  us, however temporary would be most appreciated. Thank
you. 

Regards,
Carlo Accorsi

-----Original Message-----
From: Emmanuel L├ęcharny [mailto:elecharny@gmail.com] 
Sent: Wednesday, April 11, 2012 1:17 PM
To: users@directory.apache.org
Subject: Re: Plea for help with search performance

Ok,

after some investigation, I'm pretty sure that the indexes are not created. When I create
the indexes, I get those results :

- 80000 entries injected into the server, with uid=<number>
- displayName is indexed, uid is not indexed

- first search :
connection.search("dc=example,dc=com", "(displayName=12345Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D)",
SearchScope.SUBTREE, "*" );

result :
Delta search : 16 ms
entry found :
     dn[n]: uid=12345,dc=example,dc=com
     objectclass: organizationalPerson
     objectclass: person
     objectclass: inetOrgPerson
     objectclass: top
     title: Snr Operations Technician (D)
     uid: 12345
     description: UI - S
     businesscategory: Ops MDS (Malaysia) Sdn Bhd
     departmentnumber: SMDS - UIA/G/MMO52D
     employeenumber: A-A-R.Awg-Rosli
     givenname: Awg-Abd-Rahim
     cn: Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D
     sn: Awg-Rosli
     telephonenumber: 555-1212
     mail: A-A-R.Awg-Rosli@acme.com
     displayname: 12345Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D

- Second search :
connection.search("dc=example,dc=com", "(displayName=34567*)", SearchScope.SUBTREE, "*" );

result :
Delta search substring : 9 ms
entry found :
     dn[n]: uid=34567,dc=example,dc=com
     objectclass: organizationalPerson
     objectclass: person
     objectclass: inetOrgPerson
     objectclass: top
     title: Snr Operations Technician (D)
     uid: 34567
     description: UI - S
     businesscategory: Ops MDS (Malaysia) Sdn Bhd
     departmentnumber: SMDS - UIA/G/MMO52D
     employeenumber: A-A-R.Awg-Rosli
     givenname: Awg-Abd-Rahim
     cn: Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D
     sn: Awg-Rosli
     telephonenumber: 555-1212
     mail: A-A-R.Awg-Rosli@acme.com
     displayname: 34567Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D

-third search :
connection.search("dc=example,dc=com", "(uid=67890)", SearchScope.SUBTREE, "*" );

result :
Delta search no index : 38985 ms

entry found :
     dn[n]: uid=67890,dc=example,dc=com
     objectclass: organizationalPerson
     objectclass: person
     objectclass: inetOrgPerson
     objectclass: top
     title: Snr Operations Technician (D)
     uid: 67890
     description: UI - S
     businesscategory: Ops MDS (Malaysia) Sdn Bhd
     departmentnumber: SMDS - UIA/G/MMO52D
     employeenumber: A-A-R.Awg-Rosli
     givenname: Awg-Abd-Rahim
     cn: Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D
     sn: Awg-Rosli
     telephonenumber: 555-1212
     mail: A-A-R.Awg-Rosli@acme.com
     displayname: 67890Awg-Rosli, Awg-Abd-Rahim SMDS-UIA/G/MMO52D


So if the index is properly set, the search takes milliseconds to complete (which is expected).
Otherwise, it can take tens of seconds...

Two possibilities :
- you don't initialize the index correctly. Please provide your configuration and if you are
using ADS embedded, please provide the code that you use to embed it
- we have an issue in the way we initialize the index, which is more likely to be the problem,
as I fixed one issue in this area 2 weeks ago (in M7-SNAPSHOT) and I just fixed another one
while doing the experiment...

I'd like to rule out the first hypothesis first, but in any case, I'll do some more check
tonite or tomorrow.


Many thanks for your patience.


-- 
Regards,
Cordialement,
Emmanuel L├ęcharny
www.iktek.com

Mime
View raw message