When to worry about load averages?

I recently installed nagios on one of my linux boxes (rhel 4, postfix email server), to see how it works. I regularly am getting these “critical” warnings. Should I be worried?

Example:

[quote]
Date/Time: Tue Jun 9 08:33:04 PDT 2009

Additional Info:

CRITICAL - load average: 6.36, 6.19, 4.73[/quote]

How many CPU’s/cores does this box have?

Two processors, single core.

With 2 cores I would start checking things as soon as the load avg. is more than 2 over a prolonged time. See the Wikipedia for more details on Linux load average.

top can help you determine which processes are the culprits.

[quote=“Evert”]With 2 cores I would start checking things as soon as the load avg. is more than 2 over a prolonged time. See the Wikipedia for more details on Linux load average.

top can help you determine which processes are the culprits.[/quote]

and the culprits are clamscan, MailScanner and imapd…

Since no one has complained about the email server being slow, does it really matter? It is at least one more piece of information to try and convince my boss to move to a hosted service for email, as I don’t have the time to deal with it-but I’ll now have to see if this is affecting our service in any way…

If you are not running an ISP level mail server you shouldn’t have the load that high for long periods of time.

A useful thing i did at my old place was greylisting… that really cut down on the spam filtered by the mail server, with a VERY limited amount of false positives, due to the way greylisitng works and some “mailservers” NOT retrying after a service unavailable reply… :frowning:

Greylisting is even available for postfix in debian packages with very limited work.

I’d never give up a mailserver to have the mail administered by somebody out of house… but that may be only my idea as an ex mailserver admin, it really depends on how much else work you have :slight_smile:

Hope it helps.

Luca

[quote=“luca”]If you are not running an ISP level mail server you shouldn’t have the load that high for long periods of time.

A useful thing i did at my old place was greylisting… that really cut down on the spam filtered by the mail server, with a VERY limited amount of false positives, due to the way greylisitng works and some “mailservers” NOT retrying after a service unavailable reply… :frowning:

Greylisting is even available for postfix in debian packages with very limited work.

I’d never give up a mailserver to have the mail administered by somebody out of house… but that may be only my idea as an ex mailserver admin, it really depends on how much else work you have :slight_smile:

Hope it helps.

Luca[/quote]

This is just a mailserver for a small office… I’ve noticed some lag with sending/receiving to/from external accounts sometimes-but always just a few seconds. No one else here has ever had any problems with it or even noticed the delay. I get probably around a dozen notices a day from nagios about the load average being too high…

I just looked at greylisting-it seems like it could be interesting, but we can’t have emails from new senders delayed until its resent. I have MailWatch with ClamAV and SpamAssasin-and while it filters out a lot of spam, it’s still killing this older box, while not actually blocking enough spam.

As far as giving up control of my mailserver… Well, it really wasn’t my first choice. But because we’re non-profit I can get google’s service for free. I don’t have any equipment budget this year, and I don’t have time to do everything I should with the spam problem. This gives me more space per user, frees up a server I can then use for testing/development, and will help with the spam my users are still getting. Also, it frees me up a bit, I’m the sysadmin, which means I have mail, phone, file server, ldap, apple servers, firewalls, wireless, financial servers, streaming media servers, and user support shudder.

Anyways, thanks for the info… I’ll now be trying (in the next couple weeks) to get nagios up and running on all my systems that I can…