Thursday, March 16, 2006
Technical difficulties
I've seen more MQ reason codes in four months at the new place than I did in four years at CLA.
I've been working the QA MQ cluster, getting it operational. This past week has been spent doing failover testing. I tried doing it manually (don't ask) and ended up screwing both servers. After the sysadmins fixed them up (I don't think they're too happy with me right now), I did shutdowns and reboots, and everything was fine. Shutting down one server caused all the queue managers to failover to the other as they should. "Knockdown" testing on the other hand only works on the one side of the cluster, the server with the heartbeat.pid file. I looked at the report from the IBM proof-of-concept, and it looks like they consistently used only the one queue manager for knockdown testing.
The production cluster had been set up with queue manager names and a repository name that aren't consistent with the existing convention. I'm trying to change them but so far they have resisted my efforts. I may have to delete and recreate.
D.
I've been working the QA MQ cluster, getting it operational. This past week has been spent doing failover testing. I tried doing it manually (don't ask) and ended up screwing both servers. After the sysadmins fixed them up (I don't think they're too happy with me right now), I did shutdowns and reboots, and everything was fine. Shutting down one server caused all the queue managers to failover to the other as they should. "Knockdown" testing on the other hand only works on the one side of the cluster, the server with the heartbeat.pid file. I looked at the report from the IBM proof-of-concept, and it looks like they consistently used only the one queue manager for knockdown testing.
The production cluster had been set up with queue manager names and a repository name that aren't consistent with the existing convention. I'm trying to change them but so far they have resisted my efforts. I may have to delete and recreate.
D.
