Path: csiph.com!v102.xanadu-bbs.net!xanadu-bbs.net!news.mixmin.net!rt.uk.eu.org!newsfeed.xs4all.nl!newsfeed1a.news.xs4all.nl!xs4all!newsgate.cistron.nl!newsgate.news.xs4all.nl!post.news.xs4all.nl!not-for-mail
MIME-Version: 1.0
In-Reply-To: <CACwCsY7BWeV14+=OHbgYyfwV8qGkTv_nRC4e4+Ahxv3+RP+tng@mail.gmail.com>
References: <CACwCsY7BWeV14+=OHbgYyfwV8qGkTv_nRC4e4+Ahxv3+RP+tng@mail.gmail.com>
Date: Fri, 18 Jul 2014 03:32:08 +1000
Subject: Re: Blocked thread
From: Chris Angelico <rosuav@gmail.com>
Cc: "python-list@python.org" <python-list@python.org>
Content-Type: text/plain; charset=UTF-8
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.11934.1405618331.18130.python-list@python.org>
Lines: 36
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:74651

On Fri, Jul 18, 2014 at 2:26 AM, Larry Martell <larry.martell@gmail.com> wrote:
> I have a python cx_Oracle script that does a delete from a table.
> Usually this takes well under 1 second. But sometimes it takes 1 to 2
> minutes. I wanted to monitor that delete and if it's taking too long I
> want to see what is blocking it. I run the delete sql in a thread...

I don't know Oracle specifically, but if it's anything like
PostgreSQL, you'll probably do better with a completely separate
connection to the server, which might need to be a separate process.
In PostgreSQL, I can query currently-active transactions thus:

rosuav=> select state,query from pg_stat_activity;
        state        |                   query
---------------------+-------------------------------------------
 idle in transaction | select * from pg_stat_activity;
 active              | select state,query from pg_stat_activity;
 active              | drop table test;
(3 rows)

(Better than that: Add "where pid=..." to that, using the backend PID
provided by the thread you're monitoring, by "SELECT
pg_backend_pid()". But that's even more PostgreSQL-specific.)

With info like that, you can see what's happening, and whether it's
stalled out or in a query or whatever. You should also be able to get
some timestamps (Postgres can do that, I would be highly surprised if
Oracle can't), such as when the transaction started, so you can see
how long it's been stalled.

Thing is, this requires a quite separate connection, which means
you're monitoring the far end rather than the local thread. I suspect
this will give you better results; Oracle's bound to have facilities
for doing this, whereas your local thread may or may not be usefully
monitorable.

ChrisA