Re: Oracle Management System (OEM) on linux
Date: Wed, 24 Feb 2010 09:16:51 -0600 (CST)
> We are trying to decide whether to put our oms on linux. We've always
> had it on windows so this would be new to us. Has anyone done this and
> if so how has it run? Specifically, has it been consistently good in
> running jobs and sending alerts? Any other useful info is also appreciated.
In my case, no it hasn't been consistently good, but I don't believe it's a Linux-vs-Windows issue. One of the major problems I've had (and spouted off about on this list) was scheduled jobs simply not running for no apparent reason. And since these were backup jobs, which GC does not allow making changes to, the only fix was to drop and recreate them. Patch 7576937 seemed to have corrected it, but I've moved my backup jobs out of GC.
I've also made it known my, erm, "displeasure" with GC reliability in other areas. Most recently, one metric failed to correctly report the lack of freespace in a production tablespace. Only after I added more space to the datafiles did the metric correct itself. There is a litany of articles on MOS that deal with freespace metric issues in GC. Unfortunately, none of them seemed to fit my particular situation. Needless to say -- but I'll say it anyway -- I'm in the process of replacing GC now due to reliability issues.
> I already know about the problems installing it because we have a test
> system installed. I'm looking for long term stability and reliability info.
Problems installing on Linux? Now that one's new to me. I tried for a week or two to install 10.2.0.3.0 on Windows with no success. This was during the "dead zone" of the then-new DST 2007 changes, in between the old and the new changeover dates. I convinced my SR to let me try OEL 4, and had GC up and running in a morning. Which Linux are you attempting to install it on?
I monitor just 5 databases, including the OMS repository. With all the other targets that I don't care about, this results in about 30 targets in GC. The spartan server I'm using for the OMS and repository DB has 2GB of RAM. After about two months, swap usage hits >1.5GB and I need to bounce the OMS. I have seen nothing to believe that this issue could be solved by adding RAM, only delayed. I'm using the 32-bit version of OEL 4, if that makes any difference to anyone.
Just my $.02,