debugging ldap synchronisation

[Michael Regelin]

Hello,

We are trying to synchronise Confluence 3.5.13 (linux-tomcat) with a vldap (openLdap).

our Ldap contains about 32000 users and 650 roles (groups).

When The synchronisation start. It takes more than 25 hours to finish. An exemple here where the synchroniseMembership took more than 4 hour for one group...???

2012-09-11 19:12:52,737 INFO [QuartzScheduler_Worker-10] [directory.ldap.cache.AbstractCacheRefresher] synchroniseMemberships found [ 20473 ] remote user-group memberships, [ 0 ] remote group-group memberships in [ 14164385ms ]
2012-09-11 19:13:33,127 WARN [QuartzScheduler_Worker-10] [atlassian.crowd.directory.DbCachingRemoteChangeOperations] addUserMembershipsForGroup Could not add the following missing users to group [ PRO-WIKI-INDUS-LECTEUR ]: [HEULING-ALT, VEGAMOLINAC]
2012-09-11 19:13:33,127 INFO [QuartzScheduler_Worker-10] [atlassian.crowd.directory.DbCachingRemoteChangeOperations] addUserMembershipsForGroup added [ 20471 ] user members to [ PRO-WIKI-INDUS-LECTEUR ] in [ 40374ms ]
2012-09-11 19:13:35,115 INFO [QuartzScheduler_Worker-10] [directory.ldap.cache.AbstractCacheRefresher] synchroniseMemberships found [ 1 ] remote user-group memberships, [ 0 ] remote group-group memberships in [ 1949ms ]

Does someone knows what classes to debug or what to look for in this situation.

thanks for help.

Michael

[GuilhermeA · Accepted solution]

Hello Michael,

As far as I know something that would help you on this would be adding the Crowd library to the Logs, this line should be added com.atlassian.crowd under Admin > Logging & Profiling.

However I'm not really sure if this will add everything you need. Someone else should give you a better response, but will give you a path at least.

Another thing to sum this up, you could break your LDAP into smaller pieces if you want, for Region, or for category or even for Tree structure.

Cheers,

Guilherme

[Michael Regelin]

Hi Guilherme,

sorry for being late in answering you.

Thanks for your post. I checked and mark this classes in debug:

com.atlassian.crowd.directory

com.atlassian.crowd.embbeded

It helped me a lot in finding a clue, not the solution but some clues.

one is that the virtual ldap server was really bad with his own caches. So it's better now.

thanks again.

MIchael