heavyweight OS monitoring as standard
Date: Tue, 05 May 2009 13:54:26 -0400
I'm at Collaborate right now, and we just finished the RAC SIG bird-of-a-feather session (it's a group discussion, and was quite good). Had an interesting point come up - we were discussing troubleshooting and asked how many people ran OSW as a standard practice on their clusters. (Metalink note 301137.1) I think about half the room raised their hands.
OSW is also included in RACDDT. (Metalink note 301138.1) And there's also a new tool from Oracle called IPD which seems to gather similar information into a little Berkeley DB (running processes, runqueues, etc): http://www.oracle.com/technology/products/database/clustering/ipd_download_homepage.html
How many people on this list install OSW as a standard practice on their database servers? Do you install some other heavy-weight monitoring tool as a standard? I always thought that most people only installed light-weight monitoring tools by default, would only install something like OSW and if there were problems since it captures so much data. But maybe this should become a standard practice for us; it would make troubleshooting machine crashes after the fact a *lot* easier.