innfeed - "blocked sleeping 120" errors"
Petr Novopashenniy
pety at rusnet.ru
Mon Mar 30 14:42:25 UTC 2009
On Sat, 28 Mar 2009, Julien ?LIE wrote:
JL> Hi Petr,
JL>
JL> > These innfeed processes supports ~85 peers. Yes, I have broken peers with
JL> > "rejected 502" errors, or with broken DNS..
JL>
JL> You then see "cxnsleep response unknown: 502" in your logs.
JL> innfeed goes to sleep when receiving such an answer. I assume you do not
JL> send tons of articles to those broken peers...
No. I'm trying to comply the requested subscription.
JL>
JL>
JL> > > JL> Is it happening all the day?
JL> >
JL> > Mostly after crash, when server accepts backlog from my other INN
JL> > (2.4.3) servers.
JL>
JL> Then it explains why innfeed receives a lot of articles at the same
JL> time (fast local network) and does not cope with that amount of work
JL> (though it should).
In what other cases this can happen?
In other words, better to start more innfeed processes, and not to
increase the number of peers to 100 (default maximum) for one innfeed
proc, and then add the second innfeed proc?
JL> --
JL> Julien ?LIE
JL>
Petr Novopashenniy
More information about the inn-workers
mailing list