Hi, Thanks for the reply. I don't know for sure, but it doesn't look to me like this is the issue. But it makes me think about things like that, but I am not getting any messages about file descriptors or anything in the /var/log/messages. When I do a netstat or sockstat I see hundreds, but not thousads of connections.
Thanks, Eric At 01:53 AM 3/27/03 -0500, Rudy Lippan wrote: >On Wed, 26 Mar 2003, Eric Frazier wrote: > >> We have been stuck with a CGI script that is tracking clickthrus, and my >> first impulse to deal with this problem is to just use mod_perl and a >> persistent connection with Apache::DBI, but we don't have that choice right >> now. What is happening is the script, which has to connect to a remote >> database, again, not my choice, periodicly has connection failures. But >> these failures are of a very random, but regular nature. The database >> server, load is small for what it can handle, mysql on that server is > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >How many connects/sec is the server getting? Because if your connect rate >gets high enough, you might be running out of ephemeral ports. Say you end >up getting 4000/30sec give or take. > > >> DBI->connect(affiliate_program;hostname=xxx.xxx.xxx.xxx >> ;mysql_connect_timeout=10) failed: Can't connect to MySQL server on >> 'xxx.xxx.xxx.xxx' (113) at >> /usr/local/lib/perl5/site_perl/5.6.1/mod_db_connect.pm line 57 >> can't connect: Can't connect to MySQL server on 'xxx.xxx.xxx.xxx' (113)\ >> >> That will repeat for about 30 seconds, then go away. I tried using the > >If you run through all of the epemeral ports on the client within 2*MSL, >the server will start refusing connections up until another port becomes >free. Now, IIRC, instead of using 2*MSL, FreeBSD uses a default quite >period of 30sec :) > >HTH, > >Rudy > > (250) 655 - 9513 (PST Time Zone) "Inquiry is fatal to certainty." -- Will Durant
