[prev in list] [next in list] [prev in thread] [next in thread] 

List:       qmail
Subject:    Re: qmail-send stop/start problems
From:       Kyle Wheeler <kyle-qmail () memoryhole ! net>
Date:       2007-01-29 17:08:51
Message-ID: 20070129170851.GC10817 () c-76-18-79-168 ! hsd1 ! nm ! comcast ! net
[Download RAW message or body]


On Monday, January 29 at 02:01 PM, quoth Rejaine Monteiro:
> After 10 seconds, I kill all qmail-remote and qmail-local process:
> #killall qmail-remote ; killall qmail-local ; sleep 5
> qmail-local: no process killed

Try:
    killall -9 qmail-send
    killall -9 qmail-remote
    killall -9 qmail-local

The -9 means "really, seriously, it's not allowed to finish up or shut 
down gracefully, it must die IMMEDIATELY". The only thing that can 
survive a SIGKILL (which is what -9 sends) is a process that's tied up 
in the kernel (i.e. your machine is having issues). A regular 
`killall` just sends a SIGTERM, which is more of a request than a 
command.

> svstat shows qmail-send appears to be down:
> #svstat /service/qmail-send
> /service/qmail-send: down 54 seconds, normally up

Svstat doesn't show the current state of the children of the process 
it started. The process it started was "qmail-start" - when that dies, 
its children (qmail-send, qmail-lspawn, qmail-rspawn, qmail-clean, 
etc.) can stay alive and svstat will never know.

> #But qmail-send continues running (and delivery messages):

It's log files should indicate that it is, in fact, merely *finishing* 
the delivery of messages that it started delivering before you killed 
qmail-start.

> The only way I see to stop definitively qmail-being is commenting 
> SVSCAN line in /etc/inittab, run  "init q" and later  commenting out   
> and run "init q" again to re-start all process...

That wouldn't do it.

~Kyle
-- 
It's amazing how much "mature wisdom" resembles being too tired.
                                                 -- Robert A. Heinlein

[Attachment #3 (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]