Memory Leaks

[Jerry Benton]

Hello,

Has anyone noticed memory leaks in MailScanner? After a few days of a
server that is not actually doing anything and just sitting there, memory
becomes consumed. Ideas?

[Paul Sand]

Post by Jerry Benton
Has anyone noticed memory leaks in MailScanner? After a few days of a
server that is not actually doing anything and just sitting there, memory
becomes consumed. Ideas?

Not seeing it here. But (on the other hand) this may be why this is
in MailScanner.conf:

# To avoid resource leaks, re-start periodically. Forces a re-read of all
# the configuration files too, so new updates to the bad phishing sites list
# are read frequently.
Restart Every = 14400

Is this parameter enabled for you?

[Alex Neuman]

1. Consumed by what?
2. Are you sure it's not caching and such? Ideally, a Unix system should
*not* have "unused" memory - it should use as much as possible between
system processes and the cache.



On Sun, Nov 3, 2013 at 12:53 PM, Jerry Benton

Post by Jerry Benton
Hello,
Has anyone noticed memory leaks in MailScanner? After a few days of a
server that is not actually doing anything and just sitting there, memory
becomes consumed. Ideas?
--
--
Jerry Benton
Mailborder Systems
www.mailborder.com
--
MailScanner mailing list
mailscanner at lists.mailscanner.info
http://lists.mailscanner.info/mailman/listinfo/mailscanner
Before posting, read http://wiki.mailscanner.info/posting
Support MailScanner development - buy the book off the website!

--
Alex Neuman van der Hans
Reliant Technologies / Vida Digital
http://vidadigital.com.pa/

[phone]
[phone]
[phone] (USA)

Follow @AlexNeuman on Twitter
http://facebook.com/vidadigital
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mailscanner.info/pipermail/mailscanner/attachments/20131103/0121c1dc/attachment.html

[Robert Lopez]

Jerry I have seen on very busy systems the number of MS child processes
continue to increase well beyond the Max Children and when that happens the
numbers in /proc/meminfo all start to change. At this time I am watching
them but not really understanding what is happening. I see Committed_AS
really starts to increase when the number of children increase.

As to a machine that is actually not doing anything. I have two of them
associated to back burner projects. Postfix and MailScanner are running and
the only email that is processed is a few root cron jobs per day. On them
the /proc/meminfo numbers do not noticeably change. I know leakage, if it
does exist, may come from routines not used by these two systems. Paul
makes a good point. I run with Restart Every = 7200 on all gateways.
However, when big spikes hit I find manually restarting helps.