Fellow DBAs,
I am working on a problem for classmates.com. Here is the problem.
There are several web servers connecting to the oracle database
(8.1.6.2.0). Each webserver uses apache/perl-dbi and establishes 30
persistent connections to the database. These connections are
established when the apache server is started/bounced/restarted. When
the maxrequestsperprocess (parameter in perl dbi) reaches 20000 for any
process then that process is killed (kill -9) and a new process is
established. Here process refers to the perl process which connects to
the database. There are 2 scenarios under which a connection is
established -
1. When the web server is restarted/started
2. When the process has processed 20000 requests
So given the above each web server cannot have more than 30 connections
at any point in time. However it has been found that sometimes they
have more than 30 connections per web server. The extra connections are
most likely dead connections. I say most likely because when this
problem occurs they don't call us, they bounce the database. They say
they can't wait because it is a production database. I have verified
and been told by my client that only perl dbi connects occur from the
web server. There are no parallel queries or other connections to the
datbase from the web servers besides the perl dbi connects.
I have suggested 2 things
1. Use sqlnet.expire_time = 5 or 10. This is a workaround for the dead
connections.
2. Instead of killing the perl process (see above) do a disconnect
(graceful exit from the db) since when a client (perl process) is killed it
can possibly leave dead connections in the database. This could be
the solution.
I have looked at the listener logs and done sqlnet trace on the server
and don't find anything abnormal. I have also monitored the current
connections to the database.
They will be trying both of my above suggestions.
What makes me queasy is that they claim that this did not happen in their
previous co-lo. They claim the network architecture, database version
and patch levels, apache webserver/dbi version and patch levels, os and
everything else is identical. Ofcourse here we have the f5 load alancers
which is before or above the web servers.
I have searched metalink and haven't come across anything like this
except for my solution 2 above.
Have you all seen anything like this? Any light u can throw will help.
Regards
--dj
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com
--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author: David Jones
INET: [EMAIL PROTECTED]
Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
San Diego, California -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from). You may
also send the HELP command for other information (like subscribing).
