EXIM is running rich, anyone have thoughts on how to chill it out?

[DWHS.net]

EXIM for as far back as I can remember is the heaviest resource on our servers. Anyone have thoughts on how to keep it more chill?

This is a open question for opinions and anyone that might know how to run EXIM with high mail smoothly.

It almost always is running at 99% of it's load level.

Any thoughts good or bad is much appreciated.

[chirpy]

A few important considerations if you haven't already done so:

1. Make sure all the /etc/valiases/domain.com are using :fail: and not :blackhole:
http://www.configserver.com/free/fail.html

2. Where possible, make sure that all the valiases domains have the catchall disabled and set to :fail:

3. Install a dictionary attack ACL:
http://www.configserver.com/free/eximdeny.html

4. Make sure that you have not set split_spool_directory to no (i.e. use the default yes)

5. Make sure you're running exim v4.50:

exim -bV

If it's not:

/scripts/updatenow
/scripts/eximup

6. Make sure that you don't have the queue processing running too frequently, i.e. leave WHM > Tweak Settings > Number of minutes between mail server queue runs > 60

7. Make sure that you have both of the following selected in WHM > Exim Configuration Editor>

Verify the existance of email senders
Discard emails for users who have exceeded their quota

8. In conjuction with sender verification, make sure that you're running bind on your server locally and that it is working and that your servers main IP address appears as the top nameserver in /etc/resolv.conf

9. Limit the length of time that you keep bounce emails in the queue by adding the following into the first texarea of the advanced exim editor:

timeout_frozen_after = 2d
ignore_bounce_errors_after = 12h

All the above can often make a huge difference to the exim load.

[Monil]

Hi chirpy,

I have a question for you. Currently my mail server has about 660,000 mails in the queue, most of which are bounces with frozen status.

Would using:

timeout_frozen_after = 2d
ignore_bounce_errors_after = 12h

slowly delete the frozen mails and allow my mail server to continue functioning normally without delay, etc?

Any response would be greatly appreciated.

[RickG]

Both of these threads contain tips as to how you can purge the mail in the queue:

http://forums.cpanel.net/showthread....ear+mail+queue
http://forums.cpanel.net/showthread....ear+mail+queue

[chirpy]

The problem comes when you don't actually want to purge the whole queue, but just clear off bounces which those two variables should indeed help with.

[Solokron]

How do you impliment these two features in exim config editor?


7. Make sure that you have both of the following selected in WHM > Exim Configuration Editor>

Verify the existance of email senders
Discard emails for users who have exceeded their quota

[Solokron]

Nevermind Been used to going straight to advanced.

How do you impliment these two features in exim config editor?


7. Make sure that you have both of the following selected in WHM > Exim Configuration Editor>

Verify the existance of email senders
Discard emails for users who have exceeded their quota

[DWHS.net]

This is a great topic, exim is always at the top of my cpu usage meters.

My resolve file shows this.

search dwhs.net
nameserver 66.250.110.98
nameserver 66.250.110.99

That domain is not on this server but is the second part of the host name. dwhs666.dwhs.net (evil server)

And the name servers are not the name servers for the server but for the data center.

Do you think that would matter?

Thanks,

[chirpy]

Yes, it certainly can. If your NOC's DNS servers become slow or non-responsive your server will have a lot of problems, particularly with email which can cause quite a backlog, especially if waiting for timeouts on DNS queries. I'd always recommend having a locally running DNS server, rather than using a NOC's, otherwise you're at the whim of their service - if you don't want to use it as full blown nameserver you can set one up to just be a local name resolver.

[beano]

At the end of the exim configuration editor there is a section for "retry rules."

Does anyone know whether the rules go in the first box or the second box?