Reputation: 127
I have a solution that uses 2 OpenAMs (v12.0.0) and 2 OpenDJs (v2.6.0). The OpenAMs are behind a load balancer and the OpenDJs are configured so 1 openAM will prefer 1 openDJ and the other OpenAM will prefer the other OpenDJ. Each OpenAM machine has the OpenDJ it prefers hosted on it. I followed this post to configure the OpenDJ instances in OpenAM
So once I got my OpenAM configured with two servers in the site I then looked at the value of com.iplanet.am.lbcookie.value in the site config which reportedly should give me the serverId and then calculated the siteId to be 1 more than the highest serverId. I got 01 for the serverid of my first instance and 03 for my second. So in my LDAP server values I have the following
opendj1:1389|01|04
opendj2:1389|03|04
From my reading this should mean openDJ1 will be used by openam1 (server id 01) and openDJ2 will be used by openam2 (server id 03). If OpenAM 1 can't access OpenDJ1 it will failover to OpenDJ2. Is this correct?
The reason I ask is that if I look at the OpenAM access logs I see almost a 50 - 50 split in the number of requests that each instance is handling. Though if I look in the OpenDJ access logs the requests seem to be favouring OpenDJ1 ie 75 - 25 split between OpenDJ1 and OpenDJ2.
Any advice welcome.
Upvotes: 0
Views: 572
Reputation: 2995
You should check the access logs of both DJ servers and identify which AM server is responsible for what proportion of traffic. Once you see the culprit, you should make sure that the OpenAM server in question doesn't have any connection issues mentioned in its debug logs.
There is a good chance that one of the AM servers had to failover to the other DJ instance. Please keep in mind, that after a failover and recovery, OpenAM retains the connections made during the failover and will keep on sending heartbeat requests to that DJ node.
Upvotes: 0