> On Tue, Mar 31, 2009 at 2:02 PM, Jared Still <jkstill@xxxxxxxxx> wrote: > >> What would you say are the essentials for database monitoring? > > Hard to believe that no one has any input on this. One of the perils of requesting non-critical info around Financial Month End close... :) Small-to-mid 24x7 manufacturing business. A svelt 15 in IT at this location, 25 in total. One DBA for JDEdwards on three 10.1 Oracle DBs, one non-critical 10.2 XE (APEX), one 10.2 GC repository, and a slew of SQueaL Servers. Since the Tuning and Diag packs were already purchased, I installed Grid Control 10.2.0.3.0 and do much of my automated monitoring through that. Of course, I believe there are gaps in it's monitoring that I supplement through a Perl script (very unlike your own!) that runs on each Oracle DB server. In addition to your monitoring, my script includes searching for the word "error" without case-sensitivity in every log and trace file under $ORACLE_BASE (pre-11g). It pages me with fun errors like: C:\Oracle/agent10g/sysman/log/emagent.trc : 2009-03-30 20:30:44,335 Thread-3088 ERROR : (nmecmgr.c,3205):Memory 0x0 encountered, expect struct_id=11011 GC also has metrics to monitor: -- Blocking locks (silly app locks itself up occasionally) -- "Bad" SQLs -- Disk footprint over time (custom) -- Long-running SQL (custom) I have non-GC scripts/procedures/jobs to gather and/or report information on object growth, temp usage, and security. Some are scheduled, some are on-demand. SQueaL Server *may* email me if a job fails, but only if mail's able to be setup on it (SS2K -- 2K5 actually uses SMTP). Working to shore up and centralize SQueaL Server monitoring and management now with a set of custom (read: no capital request required) scripts. That's all I can come up with on Month End... Rich -- //www.freelists.org/webpage/oracle-l