Hi
from 2.6 must-fix list
- There are those mysterious TCP hangs of established state sockets.
Someone has to get a good log in order for us to effectively debug this.
I think this is what I've seen a few times when using distcc.
Client (x.x.x.x) is running 2.5.67 and server (y.y.y.y) is running
2.4.20
There's a router in between with sometimes clogged interfaces so there
is some packetloss and the hang looks related to this.
When the hangs happen there's a lot of data sitting in send-q on the
client. The connections continues a few seconds after the timer has
expired iirc.
I've included two tcpdumps of these hangs. I've included everything from
the beginning of the connections, gzipped. It should be quite easy to
see when the hang happened, just look for a large jump in the
timestamps.
If more details are needed I can try to reproduce it again, shouldn't be
too difficult.
--
/Martin
stall-57682-anon.gz
Description: GNU Zip compressed data
stall-57798-anon.gz
Description: GNU Zip compressed data
|