Bug 34278 – Mailman init.d script needs to properly handle stale session.

• Product: Gentoo Linux
• Status: RESOLVED FIXED

[Description bugs 2003-11-24 12:05:35 UTC]

My computer died in a power outage yesterday, and I discovered mailman did not
start up correctly afterwards.  Since at the moment it calls a:
 su - mailman -c 'bin/mailmanctl start'
As long as the su succeeds, there will be no complaint.
I think the system needs to at least check for master-qrunner.pid being a
currently running process.  It should either warn, or attempt a:
 su - mailman -c 'bin/mailmanctl -s start'

Quick workaround is to have start and restart always use the -s parameter.  But
checking the pid file or the exit status of mailmanctl would be nice.
Ideally, zap would actually do something, presumably same as -s - remove the
stale lock and pid files, if any.

Reproducible: Always
Steps to Reproduce:
1.  /etc/init.d/mailman start
2. terminate mailmanctl somehow with extreme prejudice
3. call start, stop, restart.  doesn't matter.  mailman will not work again.

Actual Results:  
No mail delivery

Expected Results:  
Started mailman

Using mailman-2.1.3.ebuild

[Comment 1 Martin Holzer (RETIRED) 2004-01-19 11:23:25 UTC]

could you submit any patch ?

[Comment 2 bugs 2004-01-19 16:23:42 UTC]

Well, the quick workaround is to add -s to the call to mailmanctl in the init.d script.
That's what I'm doing on my computer and hardly seems to warrant creating a patch.
I can, though, if you'd feel more comfortable.


The more complex implementation of "zap" as well as testing for stale session on start, that I haven't done.

[Comment 3 Tamer Fahmy 2004-01-26 07:00:46 UTC]

can confirm this behaviour. just had a power outage and this kept me busy for a
while.

the simplistic workaround would be to add the -s option to mailmanctl as
kyberneticist suggested...

cheers,
  tamer.

[Comment 4 Martin Holzer (RETIRED) 2004-01-31 14:46:00 UTC]

in cvs