[DB-SIG] Webware and Oracle/high load

Marcos msanchez@grupoburke.com
13 Mar 2003 11:56:00 +0100

Somewhere there seems to deal really badly with bad SQL or closing
cursors before fetching all results.

What is the proposed method of dealing with bad SQL (besides solving it,
of course). Should I try to close the cursor?

What is the right way to discard results?

I seem to get about the same problems with DCOracle2 and cx_Oracle.

Are there any comparisons between them, in terms of functionality,
performance, etc.?

El mi=E9, 12-03-2003 a las 14:31, Marcos escribi=F3:
> We are testing cx_Oracle ant it seems to behave much better with this
> number of threads. It is finally working (I hate changing) on the same
> machine the Oracle+Pound+Apache is. Now we are getting 8
> pages/minute/Webware instance (each Webware uses up one CPU). We are now
> micro-optimizing.
> cx_Oracle has dumped core a couple of times. It is a relative problem,
> since Webware restarts in less than 4 seconds, but it always troubles
> me...
> With both modules, Webware seems to stop responding to ^C whenever an
> error happens in the oracle level; we have to -KILL it. :-\ I'll review
> whether I am closing cursors in case of an error.
> -----
> ref Webware: An application server in pure Python http://webware.sf.net/
> ----
> El lun, 10-03-2003 a las 21:25, Marcos S=E1nchez Provencio escribi=F3:
> > I am setting up a high load web app with webware using DCOracle2, but
> > seem to get strange dead times. Would cxOracle or SQLRelay behave bette=
> > for many (100-200 threads)? Any caveats about using DCOracle2 with this
> > number of threads? The docs say I can share a connection among threads,
> > but do I get separate trans per cursor?
> >=20
> > Thank you
> > --=20
> > Marcos S=E1nchez Provencio <msanchez@grupoburke.com>
> > www.burke.es
> >=20
> >=20
> > _______________________________________________
> > DB-SIG maillist  -  DB-SIG@python.org
> > http://mail.python.org/mailman/listinfo/db-sig