With the script below, you should be able to achieve comprehensive monitoring of OpenLDAP's monitor backend.

The extension script monitors the performance counters exposed by back-monitor, as well as monitoring replication status on all syncrepl databases, and can warn/error if a slave is out of sync by more than a set limit.

This is what the page for the test looks like.

Some example graphs generated by hobbit from the output of the script:

In conjunction with the default performance-related graphs generated by hobbit, this can be quite useful:

Compare the operations on the master with the load average of the slaves, and the memory use on the master.

(Yes, the production LDAP servers arewere missing the db4.2 memory leak patch that is incorporated into my OpenLDAP 2.3.38 packages)

How to get this working?

  1. Download bb-openldap.pl and save it in some useful place, such as ~hobbit/server/ext/
  2. Download ol-launch.cfg to /etc/hobbit/hobbitlaunch.d, and adjust any paths as necessary
  3. Download ol-graph.cfg to /etc/hobbit/hobbitgraph.d
  4. Add ol=ncv to TEST2RRD in hobbitserver.cfg, and a new line with NCV_ol="ReadWaiters:GAUGE,WriteWaiters:GAUGE,CurrentConnections:GAUGE"