Re: strange wait event

  • From: Mladen Gogala <mgogala@xxxxxxxxxxxxxxxxxxxx>
  • To: xwu@xxxxxxxxxxxxxxxxx
  • Date: Tue, 02 Aug 2005 09:27:38 -0400

oracle-l-bounce@xxxxxxxxxxxxx wrote:

On one of our test databases, I saw a lot of wait events called "ksu process alloc 
latch yield" which I have never seen before. The system was very slow. I could not 
find any information about it. Does anybody know it?


Mike, KSU related events are session login/logout events which can be traced by using events 10029/10030. This looks
like you are having many processes created at once. The usual cause of the latch contention is a CGI application which issues
a new login at every page and, of course, PMON terminating those sessions when the corresponding process is out. May be
you should consider using an application server or PHP5? I am a strong advocate of capital punishment for creators of
applications like that. Whoever writes a thing like that deserves to be executed by having to listen to Anna Nicole Smith
until dead.

Mladen Gogala
Oracle DBA
Ext. 121


Other related posts: