Perl and Apache locking

[dantarta]

I'm developing simple multiuser application based on execution of the
same script with different parameters. I use Apache 1.3 and Perl 5.0
with DBI and CGI libraries and connection to Oracle.
What I found is that when number of users execute the requests at the
same time, the requests are seemed to be locked. There are no locks on
DB level.
Any ideas?
Thanks

 

[A. Sinan Unur]

I'm developing simple multiuser application based on execution of the
same script with different parameters. I use Apache 1.3 and Perl 5.0
with DBI and CGI libraries and connection to Oracle.
What I found is that when number of users execute the requests at the
same time, the requests are seemed to be locked. There are no locks on
DB level.
Any ideas?

Yes. Please at least make an effort to explain what you are talking
about.

See the posting guidelines for invaluable information on how you can
help yourself and help others help you.

Maybe this will help:

http://www.stonehenge.com/merlyn/WebTechniques/col54.html

Sinan

 

[dantarta]

Nice article. The problem that it describes the way to do on purpose
something I suffer from...
Anyway, thanks.

 

[xhoster]

I'm developing simple multiuser application based on execution of the
same script with different parameters. I use Apache 1.3 and Perl 5.0
with DBI and CGI libraries and connection to Oracle.
What I found is that when number of users execute the requests at the
same time, the requests are seemed to be locked.

What are the observable facts which lead you to conclude that it seems to
be locks, as opposed to just really slow?

There are no locks on
DB level.

Of course there are. A DB without locks is not a DB, it is just a steaming
pile. If just select from the DB, there are very short lived locks, mostly
called pins or latches. If you insert, update, delete, or "select...for
update" then there are longer lived locks, the duration of which depends
by your transactions. Just because you don't create explicit locks doesn't
mean that there are no locks.

Any ideas?

Figure out what the problem is, then fix it.

If I didn't have any clue what's going on, then I'd but a "warn"
after every other perl statement. The "lock" must be occuring between the
last warning that was executed and first warning that should have been
executed but wasn't.

Xho