Issue Postfix cpu 100%. Mail handlers limit-out?

[Erwan]

Hi all,
We have a problem with a server with about 1000 emails account.
Plesk Obsidian & Postfix.
Since two hours, cpu is to 100%.

If a get the postfix queue, i have about 10 000 emails, not spam emails.
The services works but very degraded, emails arrive several tens of minutes later.

If i made a top command:
#top -c -b | head -50
top - 20:44:26 up 8:04, 2 users, load average: 94.63, 98.67, 104.33
And a lot of
7721 popuser 20 0 102252 11708 3828 R 5.0 0.1 0:00.15 /usr/bin/python2 -Estt /usr/local/psa/handlers/hooks/py-limit-out fix-sender-envelope,fix-sender-headers [email protected] [email protected]
7722 popuser 20 0 102252 11704 3828 R 5.0 0.1 0:00.15 /usr/bin/python2 -Estt /usr/local/psa/handlers/hooks/py-limit-out fix-sender-envelope,fix-sender-headers [email protected] [email protected]
....

The problem seems to be with the mail_handlers "limit-out". That's right ?
Why ? What is it for exactly?

# plesk sbin mail_handlers_control --list
.---.---.-------.--------------------------------------.-----------------.------------------.-----------------.
| E | P | prior | address | name | type | queue |
|---|---|-------|--------------------------------------|-----------------|------------------|-----------------|
| X | | 10 | all-recipients | check-quota | global | before-queue |
| X | | 10 | all-recipients | check-quota | global | before-sendmail |
| X | | 5 | all-recipients | limit-out | global | before-queue |
| X | | 5 | all-recipients | limit-out | global | before-sendmail |
'---'---'-------'--------------------------------------'-----------------'------------------'-----------------'
I disabled it temporarily:
# /usr/local/psa/admin/sbin/mail_handlers_control --disable --name=limit-out --type=global --queue=before-queue
# /usr/local/psa/admin/sbin/mail_handlers_control --disable --name=limit-out --type=global --queue=before-sendmail

So the mail queue was processed in 10 minutes and the cpu back to the normal.

But I would like to know the reason for the problem and details about "limit-out".

Thank you.

Erwan

 

[Erwan]

Anyone?
Can i have more explanations about mail_handlers "limit-out" and why there is a problem with it inour case?

 

[Nik G]

hello @Erwan
you can read our documentation about this feature.

In two words: this is the mechanism which can limit how many messages can send by each domain or even each mail user per hour.
Sending mail will be unavailable after limit being reached until new hour started.
So these limits checked each time when some mail user try to send outgoing email, for each domain.

If you not needed to limit your users with count emails (so they will be able to send as many emails as the wants) - disable this feature globally was the right solution.


Additionally I've pointed attention of RnD on performance problem regarding this, but did not get any response yet.
I'll try to keep you updated.

 

[yabado]

I am having this issue too, did anyone ever find a real fix?

High CPU processes look like below...

/usr/bin/python2 -Estt /usr/local/psa/handlers/hooks/py-limit-out fix-sender-envelope,limit,fix-sender-headers ...