Hello,

I have deployed a MMR cluster with a recent (about April) version of 389 from the CentOS 6 repository.

Following example 2 of this document, I have tried to set up a monitoring script on each node to verify that replication is correctly succeeding:
http://directory.fedoraproject.org/docs/389ds/howto/howto-replicationmonitoring.html

The monitoring command-line search usually works, but when replication is occurring it returns a false-positive for replication errors because some of the replicas are busy.

Rather than grepping out on the word “busy” which might lead us to miss the state when everything is erring out because everything is busy, I thought I should ask for recommendations on handling this.

My best idea is to run the command several times over several seconds and if it fails more than X times in a row, then issue an alert.  Of course that wouldn’t work if there was a longer-than-usual replication underway.  Is there a better way to do this?

Thank you,
Russ.