RE: Persistent library cache pin

From: Julio German Rojas Pelaez <jgrp_at_tid.es>
Date: Tue, 22 Jan 2002 09:21:14 +0100
Message-ID: <a2j7am$s9u1@tid.tid.es>

> Don't know if you are running the same version but
> we had the same problem when running 8.1.5 on Solaris 2.7.

We are using 8.0.6. I have read many inconcluded tar opened in Oracle regarding
8.1.* but seems they managed to get tired enough the customer to change version
/ forget it.

>
> First it will start with one particular SQL, but then cascade into
> every session that needs to load the same SQL. Eventually we had to
> bounce the database to avoid massive session pending. Killing the
> holding session does not work.

Yes, that is how it is working, we have many different SQL due a tool allowing the
users writing their own SQL. One exact query gets blocked then other. But bouncing the database does not solve the situation. We have checked an a full database import into another machine gets the same situation.

> One other note is, this seems to be triggered only when library
> cache becomes extremely fragmented.

It does not seems to be our case. The problem seems to be recorded into the tables. Received on Tue Jan 22 2002 - 02:21:14 CST