Re: PROBLEM: Sending mail-attachment > 45k with Netscape via sendmail hangs

Matthias Moeller (mattes@ice.robin.de)
Sat, 6 Mar 1999 22:46:29 +0100 (CET)


On Thu, 25 Feb 1999, Alan Cox wrote:

> > 16:48:20.829931 localhost.1088 > localhost.3333: . ack 28531 win 2542 <nop,nop,timestamp 313486 313486> (DF)
> > 16:48:20.831959 localhost.3333 > localhost.1088: P 28531:30116(1585) ack 1 win 31072 <nop,nop,timestamp 313486 313486> (DF)
> > 16:48:20.831959 localhost.3333 > localhost.1088: P 28531:30116(1585) ack 1 win 31072 <nop,nop,timestamp 313486 313486> (DF)
> > 16:48:20.831988 localhost.1088 > localhost.3333: . ack 30116 win 957 <nop,nop,timestamp 313486 313486> (DF)
> > 16:48:20.831988 localhost.1088 > localhost.3333: . ack 30116 win 957 <nop,nop,timestamp 313486 313486> (DF)
> >
>
> That stall is clean - the one end isnt reading data. That means either there
> is a problem where the kernel forgot to wake the other end (quite possible)
> or the user application is selecting on the wrong thing/busy/whatever

But the client is still adverising a window of 957. Should'nt the
sender send more data? If I slow down the client (calling sleep()
after every read()) the stall does not happen. The trace:

20:32:43.578795 127.0.0.1.3334 > 127.0.0.1.1031: P 25361:26946(1585) ack 1 win 31072 <nop,nop,timestamp 226861 206401> (DF)
20:32:43.578795 127.0.0.1.3334 > 127.0.0.1.1031: P 25361:26946(1585) ack 1 win 31072 <nop,nop,timestamp 226861 206401> (DF)
20:32:43.578845 127.0.0.1.1031 > 127.0.0.1.3334: . ack 26946 win 4127 <nop,nop,timestamp 226861 226861> (DF)
20:32:43.578845 127.0.0.1.1031 > 127.0.0.1.3334: . ack 26946 win 4127 <nop,nop,timestamp 226861 226861> (DF)
20:32:43.578865 127.0.0.1.3334 > 127.0.0.1.1031: P 26946:30116(3170) ack 1 win 31072 <nop,nop,timestamp 226861 226861> (DF)
20:32:43.578865 127.0.0.1.3334 > 127.0.0.1.1031: P 26946:30116(3170) ack 1 win 31072 <nop,nop,timestamp 226861 226861> (DF)
20:33:55.258828 127.0.0.1.3334 > 127.0.0.1.1031: P 26946:30116(3170) ack 1 win 31072 <nop,nop,timestamp 234029 226861> (DF)
20:33:55.258828 127.0.0.1.3334 > 127.0.0.1.1031: P 26946:30116(3170) ack 1 win 31072 <nop,nop,timestamp 234029 226861> (DF)
20:33:55.258914 127.0.0.1.1031 > 127.0.0.1.3334: . ack 30116 win 957 <nop,nop,timestamp 234029 234029> (DF)
20:33:55.258914 127.0.0.1.1031 > 127.0.0.1.3334: . ack 30116 win 957 <nop,nop,timestamp 234029 234029> (DF)
20:33:55.458837 127.0.0.1.3334 > 127.0.0.1.1031: . 30116:31073(957) ack 1 win 31072 <nop,nop,timestamp 234049 234029> (DF)
20:33:55.458837 127.0.0.1.3334 > 127.0.0.1.1031: . 30116:31073(957) ack 1 win 31072 <nop,nop,timestamp 234049 234029> (DF)
20:33:55.658784 127.0.0.1.3334 > 127.0.0.1.1031: . 30116:31073(957) ack 1 win 31072 <nop,nop,timestamp 234069 234029> (DF)

Here the sender continues to send data allthough the advertised window
was the same.

ciao Matthias

-
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/