From Maheswara.Rao@Sungardp3.com Tue, 27 Feb 2001 14:11:24 -0800 From: "Rao, Maheswara" Date: Tue, 27 Feb 2001 14:11:24 -0800 Subject: RE: MetaSlop -- More Screeching and Droning. Message-ID: MIME-Version: 1.0 Content-Type: text/plain Title: MetaSlop -- More Screeching and Droning. Mohan,   You brought laughter to the entire team here!!! Great. Keep it up.   Rao   Maheswara.Rao@SunGardP3.com   -----Original Message----- From: Mohan, Ross [mailto:MohanR@stars-smi.com] Sent: Tuesday, February 27, 2001 4:26 PM To: Multiple recipients of list ORACLE-L Subject: MetaSlop -- More Screeching and Droning.   Our own esteemed Paul Baumgartel wants to know how our equally esteemed Kevin would have handled the MetaCrack announcement.

  There are several possibilities, depending on what is ACTUALLY  going on in the Armani Sweatshop on Ellison Lane. Scenario One: ATTENTION:  Oracle Customers -- Special DBA Announcement.   PASS THIS TO YOUR DBA! No doubt you have recently seen performance on MetaLink do down like a career hooker on a Presidential candidate. There is a reason for this, and here it is: We had a cowboy DBA who stayed up too late one night reading "Oracle Latches, Waits, and Blinking Lights" by Tom Swift. (Doubtless you are aware of this classic review of oracle internals.) Unfortunately, aforementioned cowboy badly misinterpreted the settings of the _db_block_write_flush_or_scan_depth_increment_rigidity parameter and really screwed up our systems. After out "Global IT Group" got on the problem ( ok, the "other" cowboy rookie DBA we have ) he tried everything he could, and actually worsened things by setting the "_please_make_it_better" parameter to TRUE, as opposed to IMMEDIATE. As you may know, this undoc parameter only works in TRUE mode for Oracle8, and MetaLink is still running on Oracle 7.2.3.4.2. Finally, we called in someone competent from our front line support. She had actually quit, and was packing her desk in disgust (at not being allowed to properly address customer requests ever since the "iTar" process performed auto-deletes on all online requests) but was stopped by an alert middle manager for "one last favor". After she set the "_divert_funds_from_marketing" parameter to TRUE, everything came back for a handful of minutes. During this brief operational period, we had a "staggering" number of requests hit our front end webservers ( Apache on Linux-based Palm Pilots ) and MetaLink could not respond to the seven requests simultaneously. We are currently throwing hardware at the problem, and hope to be able to provide roughly the same level of service you'd expect at a local Seven-Eleven any day now. So Long, and Thanks for All the Power Units!!!! ################   END METALINK MESSAGE #######################################   There are other scenarios, but my hourglass just returned from my metalink search, so i've got to go now.........