Very large innfeed processes

Miquel van Smoorenburg list-inn-workers at news.cistron.nl
Mon Nov 29 11:12:12 UTC 1999


In article <cistron.ylemdbnwmq.fsf at windlord.stanford.edu>,
Russ Allbery  <rra at stanford.edu> wrote:
>Miquel van Smoorenburg <list-inn-workers at news.cistron.nl> writes:
>
>> I have 5 innfeed processes running. One of them is always getting very
>> large pretty quickly.
>
>Is that by any chance one that's handling an unusually large number of
>binaries?  I was seeing this too with my innfeed process that was handling
>large articles until I set max-queue-size for that innfeed to 5.  You
>don't want to stream binary articles too much; it takes too long to send
>them and you end up with excessively large queues.

I followed yur advice and set max-queue-size to 5. I also turned on
mmap() for innfeed, that should also prevent some more copying. Alas,
it is still growing bigger and bigger:

wormhole% ps uw 12744
USER       PID %CPU %MEM   VSZ  RSS TTY      STAT START   TIME COMMAND
news     12744 12.1 10.7 2084368 27736 ?     SN   Nov28  90:25 /usr/local/news/bin/innfeed -y -c innfeed2.conf -p innfeed2.pid -S /v

Almost 2 GB used now .. when it reaches that it will probably crash.
Weird thing is that RSS is "only" 27M, somehow 2054212K has been
mmap()ed anonymously (output from /proc/12744/maps on linux, prefixed
by size):

2054212K 429ef000-c0000000 rwxp 829f0000 00:00 0

Weird.. wondering if this is a bug in innfeed or in Linux's glibc ..

Mike.
-- 
First things first, but not necessarily in that order.
-- 
The From: and Reply-To: addresses are internal news2mail gateway addresses.
Reply to the list or to miquels at cistron.nl (Miquel van Smoorenburg)


More information about the inn-workers mailing list