heads up on em grid upgrade

From: Niall Litchfield <niall.litchfield_at_gmail.com>
Date: Fri, 12 Feb 2010 22:52:22 +0000
Message-ID: <7765c8971002121452x479b536fhbfa64c195c845b4c_at_mail.gmail.com>



Just a heads up for anyone tempted to upgrade to GC 10.2.0.5 and migrating to an 11g db as well. We did this and found that apparently agent communication didn't work properly whereas it had previously. In particular GC didn't notice when an agent went down and then manual investigation of the agent page in gc displayed the "cannot communicate with agent host" message. communication was in fact fine.

I spent a lot of time, too much time, on diagnosing comms problems with Oracle support some their problem, some mine.

It turns out that the agent unreachable alert (and therefore notification) that we should have got is triggered by an EM dbms_job (EM_PING.MARK_NODE_STATUS()
). There are various bugs/notes about job_queue processes and em jobs not being submitted. In our case none matched. However I eventually discovered that the new scheduler had been disabled in the repository. This has the unexpected side effect (well unexpected to me) of also disabling *all*dbms_jobs. Hopefully it won't take you several days to pick this up if it
affects you. I guess there will also be other cases where the new scheduler gets disabled and breaks em.

-- 
Niall Litchfield
Oracle DBA
http://www.orawin.info

--
http://www.freelists.org/webpage/oracle-l
Received on Fri Feb 12 2010 - 16:52:22 CST

Original text of this message