Re: Postgres 8.0/Windows 2000 Load testing

The problem is not only with Postgres 8.0,. It happens with 7.2.x,(cygipc),
7.4.x(cygserver) version as well. But with the previous versions atleast
the test was able to last a for 5-6 concurrent connections. But with 8.0,
with 3-4 concurrent connection itself it starts to break.

On Linux(with 7.2.1), for about 10 concurrent connections everything is
fine. After that some 'deadlock detected' exception is thrown by the
database and the connection is closed. But the database itself is responsive
and does not hang or crash.

But on windows the database fails to respond and windows itself freezes
after some time.

Thanks,
-Sarva

>Bellan Saravanan wrote:
> > While performing Load testing using DOTS
> > http://ltp.sourceforge.net/dotshowto.php ,
> > during the tests, an update to a specific table starts hanging.
> >
> > The test name is called BTCJ2
> > http://ltp.sourceforge.net/dotshowto.php#SEC26,
> >
> > "This test case mainly uses SQL commands to execute database operations
>such
> > as insert, update, select and delete. This test case uses tables BASIC1,
> > BASIC2, BASIC3. For details about the tables, refer to Appendix A."
> >
> > Postgres itself was running fine and accepting new connections and able
>to
> > perform queries on the same table. Only the UPDATE was hanging.
> >
> > There was no interesting messages in the log file. I was running with
>debug
> > level 2. I can see the last line which prints the UPDATE statement.
>Using
> > debug level 3 the file grows very large and it very slow.
> >
> > Let me know if you want anymore information.
>
>Any chance of testing this on Unix? It might be some way our MVCC is
>interacting with the test and not a Win32-specific issue.
>
>--
> Bruce Momjian | http://candle.pha.pa.us
> pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
> + If your life is a hard drive, | 13 Roberts Road
> + Christ can be your backup. | Newtown Square, Pennsylvania
>19073