TCP lockups in 2.0 ?

From: Bjorn Wesen (bjorn@sparta.lu.se)
Date: Mon Apr 17 2000 - 09:23:21 EST


Hi, I'm debugging a server app which shows a weird problem - after sending
about 2-3 gigs of data on a TCP connection, the box just stops sending
more, select() does not return "its ok to write", and if you try to write
despite this, you get EAGAIN. I'm wondering if this is a known problem in
the 2.0 kernel series (this is on 2.0.38, but was same on 2.0.33).

tcpdump shows that data packets stop coming from the box, but the box
keeps replying to the clients ack's (the client tries to read more). No
actual data is sent though..

using the ioctl to check the socket TCP send buffer, shows it had data but
then it decreases to 0 and select() still does not return "ok to write
more".

TCP_NODELAY is active on the socket.

i'm grateful for any hints. after all it might be the application which is
buggy somehow, but i thought i'd shoot this question first anyway.

/bjorn

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sun Apr 23 2000 - 21:00:10 EST