Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> ksu process alloc latch yield

ksu process alloc latch yield

From: Thorsten Jens <thojens_at_gmx.de>
Date: 10 Mar 2006 04:39:57 -0800
Message-ID: <1141994397.285975.24800@p10g2000cwp.googlegroups.com>


Hi,

I have a problem with an application based on Oracle 9i. On startup, this application reads and caches a lot of tables. The time for loading has increased substantially over the last weeks, from about 15 minutes up to 50 minutes.

When looking a the session waits with TOAD (of a session that took ~25 minutes to load the data), I got something like the following:

EVENT:          ksu process alloc latch yield
TOTAL_WAITS:     6560

TOTAL_TIMEOUTS: 6534
TIME_WAITED: 68743 EVENT: SQL*Net message from client TOTAL_WAITS: 167926
TOTAL_TIMEOUTS:      0
TIME_WAITED:     36981

[...]

I assume this means that of ~1500 seconds, ~687 seconds are spent waiting for "ksu process alloc latch yield" (mostly timing out), and another ~369 seconds are spent waiting for the network. Am I right?

What is that latch yield thing? I couldn't find too much about it, the only thing I found mentioned frequent logons/logoffs. But "logons cumulative" for that session is 1!

Database is "Oracle9i Enterprise Edition Release 9.2.0.7.0 - 64bit Production" on a 2 CPU SunFire 3800, 64bit, Solaris 8. CPU load during the time of that session at ~1.8, while the session never used more that 4% CPU.

Any ideas?

Thorsten Received on Fri Mar 10 2006 - 06:39:57 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US