[Dovecot] Fastest restart of dovecot after upgrade: HOW?:
I'm running dovecot under runit:
#!/bin/sh echo '/var/core/%p' > /proc/sys/kernel/core_pattern ulimit -n 8096 ulimit -c unlimited exec /usr/local/sbin/dovecot -F
(that way, together with incron, I can automatically generate the coredumps I'm posting all the time)
If I apply a patch and recompile, what is the fastest way of restarting dovecot?
Which part of the process structure can be killed and takes all it's children with it?
├─runsvdir─┬─runsv───ntpd
│ ├─runsv─┬─dovecot─┬─2*[dovecot-auth]
│ │ │ ├─1203*[imap]
│ │ │ ├─64*[imap-login]
│ │ │ ├─64*[managesieve-log]
│ │ │ └─64*[pop3-login]
│ │ └─svlogdkillall dovecot? Is that guaranteed to wipe out ALL child processes as well?
I had to use (at least in the past):
sv d /var/service/dovecot killall dovecot killall -9 imap-login ps auxwww|grep dovecot
to make sure everything would be dead...
-- Ralf Hildebrandt Geschäftsbereich IT | Abteilung Netzwerk Charité - Universitätsmedizin Berlin Campus Benjamin Franklin Hindenburgdamm 30 | D-12203 Berlin Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962 ralf.hildebrandt@charite.de | http://www.charite.de
On 21.1.2010, at 16.35, Ralf Hildebrandt wrote:
Which part of the process structure can be killed and takes all it's children with it?
├─runsvdir─┬─runsv───ntpd │ ├─runsv─┬─dovecot─┬─2*[dovecot-auth] │ │ │ ├─1203*[imap] │ │ │ ├─64*[imap-login] │ │ │ ├─64*[managesieve-log] │ │ │ └─64*[pop3-login] │ │ └─svlogdkillall dovecot? Is that guaranteed to wipe out ALL child processes as well?
If you set shutdown_clients=yes. Although even with =no it should stop listening.
I had to use (at least in the past):
sv d /var/service/dovecot killall dovecot killall -9 imap-login ps auxwww|grep dovecot
to make sure everything would be dead...
I guess there can be bugs.. But in general the processes should notice immediately that master died, and immediately close their listeners and optionally shut down.
- Timo Sirainen <tss@iki.fi>:
If you set shutdown_clients=yes. Although even with =no it should stop listening.
shutdown_clients=yes is the default? I think it is :)
I guess there can be bugs.. But in general the processes should notice immediately that master died, and immediately close their listeners and optionally shut down.
In some cases that didn't happen with the imap processes. Anyway. I'll just try it now.
-- Ralf Hildebrandt Geschäftsbereich IT | Abteilung Netzwerk Charité - Universitätsmedizin Berlin Campus Benjamin Franklin Hindenburgdamm 30 | D-12203 Berlin Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962 ralf.hildebrandt@charite.de | http://www.charite.de
participants (2)
-
Ralf Hildebrandt
-
Timo Sirainen