RE: Metalink forums - effects of the outsourcing

>> If you are the analyst (or the DBA) having to deal with a support
issue that doesn't have a test case is an absolute waste of time - for both parties involved. <<

I will disagree that a repeatable test case is necessary for the majority of iTAR requests in that for the great majority of times I have entered an iTAR there is an entry in the alert log that references a trace file. With the message and the trace file support should be able to identify if the issue we have encountered is a known bug, if there is a patch or workarounds, and provide advice on how to deal with the issue.

Repeatable test cases are nice but the majority of serious problems were have encountered are intermittent and are apparently not tied to any specific process but rather are due to fundamental flaws in the Oracle kernel design and code. Even with repeated system state dumps, heap dumps, and trace files created from event settings Oracle support has been unable to find the problem code for what they have acknowledged as a bug. In cases like this support just tells you to open a new iTAR after you upgrade. So we upgrade, the error re-occurs, we open a new iTAR, and the process repeats until this problem disappears and a new one takes it place on the heap.

The real problem is too much emphasis on new features and not enough emphasis on design analysis, code analysis, unit testing, and system testing.

IMHO -- Mark D Powell --

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Paul Drake Sent: Wednesday, June 22, 2005 5:01 PM
To: jrsmiley_at_gmail.com
Cc: Oracle-L (E-mail)
Subject: Re: Metalink forums - effects of the outsourcing

If you are the analyst (or the DBA) having to deal with a support issue that doesn't have a test case is an absolute waste of time - for both parties involved. I'll bet that most times, by the time the test case is constructed and entered into whatever ticketing system, that most issues are solved.

Solving solved problems sucks.

I realize that solving problems of human stupidity is more of less job security but its still highly uninterested for anything other than making fun of people backchannel. Often times, when even just glancing at a good test case, the solution jumps right out off the screen and whaps you upside the head.

I don't have a magic 8-ball or a OUIJA board here - and I like to save my guessing for opponents hands playing Tejas Hold-Em. Diagnosing "the system is slow" issues makes me want to put the carpenter bags back on. I'll leave out the Soldier of Fortune references for now.

Paul

--
http://www.freelists.org/webpage/oracle-l
--
http://www.freelists.org/webpage/oracle-l