[Mailman-Users] mailman 2.1b3 qrunner performance (backlog)
Andrew D. Clark
andrew.clark at ucsb.edu
Tue Oct 22 20:53:28 CEST 2002
To quote from bin/qrunner:
"Each named runner class is run in round-robin fashion. In other words, the
first named runner is run to consume all the files currently in its
directory. When that qrunner is done, the next one is run to consume all
files in /its/ directory, and so on. The number of total iterations can be
given on the command line."
So, if I understand this correctly, if ArchRunner is the first named
qrunner (the default), it will have to complete its run before an other
qrunner will be able to process their queues. Is this correct? It seems a
little backwards, since I'd guess that archives are usually the lowest
priority queue. Can this ordering be changed by altering the QRUNNERS = 
So, the qrunners _aren't_ really executing in parallel?
Campus Network Programmer
Office of Information Technology
University of California, Santa Barbara
andrew.clark at ucsb.edu (805) 893-5311
--On Saturday, October 19, 2002 21:55:35 -0400 "Barry A. Warsaw"
<barry at python.org> wrote:
>>>>>> "ADC" == Andrew D Clark <andrew.clark at ucsb.edu> writes:
> ADC> I'm pretty sure lag in archiving is what was causing the
> ADC> whole thing to back up in 2.0.12. So, although the backlog
> ADC> is now mostly isolated to archiving (a very good thing) I'm
> ADC> looking for a solution to fix that. Not archiving is not an
> ADC> option. Should forking more ArchRunners help with this, or
> ADC> will this just cause more contention?
> Because of the list locks, it'll only help if you're archiving
> messages for a bunch of lists. But if that's the case, it'll
> definitely help (at the cost of more cpu) because ArchRunner --
> actually all the qrunners -- can only handle one message at a time.
> Mailman-Users mailing list
> Mailman-Users at python.org
> Mailman FAQ: http://www.python.org/cgi-bin/faqw-mm.py
> Searchable Archives:
More information about the Mailman-Users