Hi,
The qrunner script in mailman started throwing up errors on our local server; copying an entry from the logs that repeats every minute when qrunner is invoked by cron.
The mailing lists had been working fine until Aug 19 3:38pm, when this problem appeared in the logs. Since then, a lot of messages are queuing up on the various lists ... please help!
Aug 19 15:38:00 2002 qrunner(25894): Traceback (innermost last): Aug 19 15:38:00 2002 qrunner(25894): File "/home/mailman/cron/qrunner", line 283, in ? Aug 19 15:38:00 2002 qrunner(25894): kids = main(lock) Aug 19 15:38:00 2002 qrunner(25894): File "/home/mailman/cron/qrunner", line 253, in main Aug 19 15:38:00 2002 qrunner(25894): keepqueued = dispose_message(mlist, msg, msgdata) Aug 19 15:38:00 2002 qrunner(25894): File "/home/mailman/cron/qrunner", line 157, in dispose_message Aug 19 15:38:00 2002 qrunner(25894): mlist.ParseMailCommands(msg) Aug 19 15:38:00 2002 qrunner(25894): File "/home/mailman/Mailman/MailCommandHandler.py", line 123, in ParseMailCommands Aug 19 15:38:00 2002 qrunner(25894): precedence = msg.get('precedence', '').lower() Aug 19 15:38:00 2002 qrunner(25894): AttributeError : 'string' object has no attribute 'lower'
On Tue, 2002-08-20 at 03:37, Sameer D. Sahasrabuddhe wrote:
The mailing lists had been working fine until Aug 19 3:38pm, when this problem appeared in the logs. Since then, a lot of messages are queuing up on the various lists ... please help!
Closer investigation revealed that only the confirmation mails from new subscribers cause this problem. If I remove them from the queue, all other pending messages go out just fine.
Does anyone know what could be the problem? One issue I can think of is that the mailing lists are of the form something@it, while the actual destination is something@cygnus.it ... could there be issues involved in this?
Sameer.
On 20 Aug 2002, Sameer D. Sahasrabuddhe wrote:
Does anyone know what could be the problem? One issue I can think of is that the mailing lists are of the form something@it, while the actual destination is something@cygnus.it ... could there be issues
DNS lookups? Do you have address canonification turned on?
On Tue, 2002-08-20 at 11:30, Sameer D. Sahasrabuddhe wrote:
On Tue, 2002-08-20 at 03:37, Sameer D. Sahasrabuddhe wrote:
The mailing lists had been working fine until Aug 19 3:38pm, when this problem appeared in the logs. Since then, a lot of messages are queuing up on the various lists ... please help!
Closer investigation revealed that only the confirmation mails from new subscribers cause this problem. If I remove them from the queue, all other pending messages go out just fine.
A search on the mailman users list solved the problem. It was an incompatibility with python, that had crept into mailman-2.0.12
Two ways to solve that - upraged to mailman-2.0.13 or upgrade python. But it turns out that python is installed in two versions on RHL 7.3, and the newer package has a binary called "python2". Renaming that to "python" was too risky, but upgrading mailman solved the problem.
Sameer.
On 20 Aug 2002, Sameer D. Sahasrabuddhe wrote:
Two ways to solve that - upraged to mailman-2.0.13 or upgrade python. But it turns out that python is installed in two versions on RHL 7.3, and the newer package has a binary called "python2". Renaming that to "python" was too risky, but upgrading mailman solved the problem.
The python 2 binary is called python2. This is so because there are incompatible changes between 1.5 and 2 which would break programs. This is also the reason why you'd have both installed. Too many people make the mistake of renaming python2 to python only to find that several seemingly unrelated things stop working at some point (alchemist for example).
Easier would be to replace #!/usr/bin/python to #!/usr/bin/python2 like this:
perl -pie 's|usr/bin/python\b|usr/bin/python2|g' *.py