Speed Up

Binary patching is slow because of the inventory

The headline of this blog post may sound a bit cryptic: Binary patching is slow because of the inventory. But actually this seems to be one of the main causes when your binary patching gets slower and slower with every patch you apply.

The more you patch, the worse it gets

Well, this is a tough headline. But you may have read my blog posts about my quarterly patching experience. It actually started with the July 2021 RU in my 19c environment when I wrote:

At least in my environment, both checks with the

Continue reading...

Speed up Upgrade Phase 65 with a new catuposb.sql

Credits to Chris Smids from Proximus in Belgium 🙂 Thanks, Chris!!!


Upgrade to Oracle 12.1.0.2 is slow in phase: #65 ?

You are wondering why phase: #65 of the database upgrade to Oracle Database 12.1.0.2 takes quite a while. You dig down into the catupgrd0.log and recognized this statement taking a while:

dbms_output.put_line('catuposb, update 4 - rows updated ' || rows_updated);
END; -- end of update for system internally generated objs
/

The cause for this issue is buried in the script catuposb.sql hitting stale histograms which did not get refreshed even if you gathered dictionary stats before the upgrade …

Continue reading...

Will gathering fixed object stats reduce recompilation time post upgrade?

Interesting question, isn’t it?
Will the time to recompile invalid objects post upgrade decreased once fixed object stats have been gathered?

First of all fixed object stats on X$-tables won’t be gathered by default [This will change with Oracle 12c where it is part of the Auto Stats Gathering job].
X$ structures are undocumented. V$ views are build on top of them and
should only be used even though it might be sometimes useful to access
X$ tables such as X$BH (buffer headers – contains information describing the current contents of a piece of the buffer cache

Continue reading...