|Summary:||mail-filter/postgrey init script fails startup under sys-apps/openrc-0.8.2-r1|
|Product:||Gentoo Linux||Reporter:||Giampaolo Tomassoni <giampaolo>|
|Component:||[OLD] Server||Assignee:||Andreis Vinogradovs ( slepnoga ) <andreis.vinogradovs>|
|Severity:||normal||CC:||hasufell, lists, net-mail+disabled, proxy-maint, qnikst|
|Package list:||Runtime testing required:||---|
|Attachments:||A patch to the /etc/init.d/postgrey init script to fix this|
Description Giampaolo Tomassoni 2011-05-29 09:45:48 UTC
Created attachment 275005 [details, diff] A patch to the /etc/init.d/postgrey init script to fix this Starting postgrey on a system working in a recent openrc model, result in the following error: /etc/init.d/postgrey: start-stop-daemon: caught an interrupt /etc/init.d/postgrey: start-stop-daemon: /usr/sbin/postgrey died /etc/init.d/postgrey: ERROR: postgrey failed to start However, the process is correctly started and running. The reason for this seems related to the fact that the start-stop-daemon function can't find the process' pid file after starting it. This is due to two fact: 1) the init script starts the postgrey daemon invoking 'start-stop-daemon --background', while the daemon is instead able to fork and detach by itself, thereby the start-stop-daemon is waiting for the wrong child; 2) the postgrey daemon is quite slow in creating/updating its pid file, such that start-stop-daemon may occasionally fail even when the --background option is removed from its runstring. Attached to this you'll find a patch to the /etc/init.d/postgrey init script, which relies on the daemon's own capability to fork and detach, and which uses the --wait option to wait for a valid pid file to be created. Please note that the --wait option in start-stop-daemon seems to be introduced by some recent version of the script, so this init script may not work on baselayout-1 systems. Also, the patch works around a (to me) unusual way of fixing the /var/spool/postfix/private/postgrey socket permissions, which was previously made by forking a sh job.
Comment 1 Eray Aslan 2011-06-08 14:20:02 UTC
I cannot reproduce this with openrc-0.8.2-r1 (even under load) and I'd like to stay backward compatible with baselayout-1 if possible (even though the init script is too hacky for my taste as well).
Comment 2 Stefan G. Weichinger 2014-08-12 18:33:41 UTC
I get issues with the pidfile as well. Will report as soon as I sorted it out. Mentioned patch does not work for me (openrc 0.12.4)
Comment 3 Julian Ospald 2014-08-13 13:22:14 UTC
is the maintainer still active?
Comment 5 Stefan G. Weichinger 2014-08-13 14:51:21 UTC
postgrey 1.35 works for me
Comment 6 Julian Ospald 2014-08-13 15:05:14 UTC
(In reply to Stefan G. Weichinger from comment #5) > postgrey 1.35 works for me which one? postgrey-1.35 or postgrey-1.35-r1? Did you test postgrey-1.34-r3?
Comment 7 Stefan G. Weichinger 2014-08-14 08:37:32 UTC
(In reply to Julian Ospald (hasufell) from comment #6) > (In reply to Stefan G. Weichinger from comment #5) > > postgrey 1.35 works for me > > which one? > > postgrey-1.35 or postgrey-1.35-r1? Did you test postgrey-1.34-r3? 1.35 1.34.r3 wasn't there back then.
Comment 8 Julian Ospald 2014-08-14 11:00:28 UTC
(In reply to Stefan G. Weichinger from comment #7) > (In reply to Julian Ospald (hasufell) from comment #6) > > (In reply to Stefan G. Weichinger from comment #5) > > > postgrey 1.35 works for me > > > > which one? > > > > postgrey-1.35 or postgrey-1.35-r1? Did you test postgrey-1.34-r3? > > 1.35 > > 1.34.r3 wasn't there back then. again: Did you test postgrey-1.34-r3?