[Mailman-Users] Excessive CPU usage [Looping read/close/mmap] (.db , .pck) (qfiles/out)
foobar at intter.net
Thu Jun 19 12:28:40 CEST 2003
> >= 0x40227000
> >read(9, "(cMailman.Message\nMessage\nq\1oq\2}"..., 4096) = 4096
> >Could you tell what is wrong, no messages are going currently out or in,
> >MTA is working properly and bounces were handled correctly.
> > PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
> >xxxxx mailman 25 0 6188 4848 2820 R 94,5 1,8 1192m python
> >What _could_ cause this kind of looping?
> >Mailman (see directory) version is 2.1.2
> >Could it be error 450 (I reject every delivery to non-existent domain):
> >XXX XX XX:XX:XX 2003 (XXYYZZ) delivery to ff at ff failed with
> >code 450: <ff at ff>: Recipient address rejected: Domain not
> Surely 450 indicates a transient failure which effectively invites the
> sender to try again.
> Maybe, if you return a 550 indicating permanent failure, Mailman will stop
> trying so hard.
By that way I get troubles if there is temporary error with A/MX-records
and if I throw away reject_unknown_recipient_domain users will not notice
immediately that recipient is not correct :/
Is there any way that mailman could delete these mails when error happens
after 5days and not re-sending in so high frequency.. this looping is
kinda annoying (banging head against wall). Yes I could reply with 550 and
mailman would delete these mails but what if we have temporarily A/MX
records broken just and those would be fixed in few hours?
I just wonder why that process loops with read/close/mmap when there is
actually no sending to be done (infinite read/close/open/rename/mmap-loop)
while sending happens to be just every 4minutes or so.
Veli Pirttila | ++Titus
More information about the Mailman-Users