Bug 12755

Summary:	systemctl reload amavisd causes daemon to crash
Product:	Mageia	Reporter:	Daniel McDonald <Dan.mcdonald>
Component:	RPM Packages	Assignee:	Thomas Spuhler <thomas>
Status:	RESOLVED INVALID	QA Contact:
Severity:	critical
Priority:	Normal	CC:	mageia
Version:	4	Keywords:	Triaged
Target Milestone:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Source RPM:	amavisd-new-2.8.2-0.rc1.4.mga4.src.rpm	CVE:
Status comment:

Description Daniel McDonald 2014-02-13 16:02:29 CET

Description of problem:
Each morning when the /usr/share/spamassassin/sa-update.cron script is run, one step is to issue this command:
	 [ -f /run/amavis/amavis.pid ] && /usr/bin/systemctl reload amavisd  >& /dev/null
The result is that the amavisd daemon crashes

Version-Release number of selected component (if applicable):
2.8.2-0.rc1.4

How reproducible:

Every time

Steps to Reproduce:
1. $ sudo systemctl reload amavisd
2. $ sudo systemctl status amavisd
amavisd.service - AMAVIS interface between MTA and content checkers
   Loaded: loaded (/usr/lib/systemd/system/amavisd.service; enabled)
   Active: failed (Result: exit-code) since Thu 2014-02-13 08:52:03 CST; 1s ago
  Process: 585 ExecReload=/usr/sbin/amavisd -c /etc/amavisd/amavisd.conf -P /run/amavis/amavis.pid reload (code=exited, status=0/SUCCESS)
  Process: 32562 ExecStart=/usr/sbin/amavisd -c /etc/amavisd/amavisd.conf -P /run/amavis/amavis.pid (code=exited, status=0/SUCCESS)
 Main PID: 32563 (code=exited, status=13)

Feb 13 08:52:03 ns1.austinnetworkdesign.com amavis[585]: Valid PID file (younger than sys uptime 4 13:34:00)
Feb 13 08:52:03 ns1.austinnetworkdesign.com amavis[585]: Signalling a SIGHUP to a running daemon [32563]
Feb 13 08:52:03 ns1.austinnetworkdesign.com amavis[32563]: Net::Server: Received a SIG HUP
Feb 13 08:52:03 ns1.austinnetworkdesign.com amavisd[585]: Signalling a SIGHUP to a running daemon [32563]
Feb 13 08:52:03 ns1.austinnetworkdesign.com amavis[32563]: Net::Server: 2014/02/13-08:52:03 Server closing!
Feb 13 08:52:03 ns1.austinnetworkdesign.com amavis[32563]: Net::Server: Sending children hup signal
Feb 13 08:52:03 ns1.austinnetworkdesign.com amavis[32563]: (!)Net::Server: 2014/02/13-08:52:03 Re-exec server during HUP
Feb 13 08:52:03 ns1.austinnetworkdesign.com systemd[1]: Reloaded AMAVIS interface between MTA and content checkers.
Feb 13 08:52:03 ns1.austinnetworkdesign.com systemd[1]: amavisd.service: main process exited, code=exited, status=13/n/a
Feb 13 08:52:03 ns1.austinnetworkdesign.com systemd[1]: Unit amavisd.service entered failed state.


Reproducible: 

Steps to Reproduce:

Daniel McDonald 2014-02-13 16:24:06 CET

Summary: systemdctl reload amavisd causes daemon to crash => systemctl reload amavisd causes daemon to crash

Manuel Hiebel 2014-02-16 18:32:03 CET

Keywords: (none) => Triaged
Assignee: bugsquad => thomas

Comment 1 Colin Guthrie 2014-02-22 12:12:33 CET

Yeah I am seeing this problem too. Will try and work out why.

CC: (none) => mageia

Comment 2 Colin Guthrie 2014-02-22 12:37:24 CET

Got it.

The /etc/amavisd/amavisd.conf file needs to be group owned+readable by amavis user. Out of the box, this is how the RPM is configured, but my system got out of sync at some point over the years.

I got the hint from this post: http://lists.amavis.org/pipermail/amavis-users/2012-July/001738.html

After correcting this, the reload works fine.

Status: NEW => RESOLVED
Resolution: (none) => INVALID

Comment 3 Colin Guthrie 2014-02-22 13:13:57 CET

I've closed this bug but I've also added a workaround in the package and submitted it to updates/testing. All it does is do a chgrp+chmod in the %post - nothing too complex!

As we've got an update pending already for MGA3, I've (currently) bundled it in there so the update will come through bug #11246 (unless QA want to keep it separate in which case this bug will likely be reopened).

All the best.