Schedule synchronise via quartz or cron

[EddieW]

So we rely heavily on Crowd's LDAP caching since our employee directory is just shy of 40,000 employees.

We re-sync every 24 hours to correspond to HR changes. The trouble is that according to my experience and documentation you can only schedule the intenrval, and not the actual time.

This means that if I start a sync now, at 12:00pm (noon) it will re-sync every day at noon. And syncing requires quite a bit of memory and CPU (75% of one vcpu)

So rather then starting the first sync manually at midnight, I would like something similar to jira's use of the quartz scheduler to specify a time, and not an interval.

Anyone found a way around this?

[Daniel Borcherding]

Have you looked into setting up a delegated connection from Crowd to LDAP.

https://confluence.atlassian.com/display/CROWD/Configuring+a+Delegated+Authentication+Directory

This will eliminate those massive performance hits during a sync. It will instead be smaller incremental loads against your LDAP server.

Also what version of crowd are you using? There have been more recent releases that address this synching issue. Perhaps an upgrade would be able to address your performance concerns.

[EddieW]

We looked at using the delegated directories but there was a limitation regarding how users are added. OUr business partners want to be able to @Mention or assign employees who may not have logged into a tool liek jIRA yet. Since we getn new employees contractors added on a regular basis this would be require a manual effort to perform manual imports on a regular basis.

[Daniel Borcherding]

Eddie,

That is an excellent reason not to use a delegated connector. Users will not be added to you instnace until they log in for the first time using this connector.

[EddieW]

we've resulted to a cron job that makes a curl call to the systems providing auth tokens as cookies to get websudo access needed to kick it off.

[Tim]

I have exactly the same problem with Confluence 4.3.3, any news on this topic? Atlassians, what about a workaround? Please focus more on enterprise features as you claim that Confluence is "Enterprise ready"!