zero window backoff question

From: Don Howard (dhoward@redhat.com)
Date: Mon May 19 2003 - 20:17:34 EST


How long does linux wait before beginning a zero window probe?

>From the trace below, it appears to wait for > 1 sec. The non-linux
client retransmits, adding a FIN flag (is adding a flag when
retransmitting reasonable?), but the linux box has already seen the
sequence number and discards it, along with the FIN. We're left with our
linux box doing zero-window probes, while the client box is in FIN_WAIT_2
waiting for a FIN that won't be comming.

Is that what I'm seeing?
Is the window probe timeout tunable?

No. Time Source Destination Protocol Info
  1 0.000000 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020120 Ack=2485442140 Win=7920 Len=0
  2 0.000411 10.5.228.15 10.12.22.218 TCP 12049 > 8887 [ACK] Seq=2485442140 Ack=47020121 Win=0 Len=0
  3 1.137483 10.5.228.15 10.12.22.218 TCP 12049 > 8887 [FIN, ACK] Seq=2485442140 Ack=47020121 Win=0 Len=0
  4 1.137819 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020121 Ack=2485442141 Win=7920 Len=0
  5 2.983763 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020120 Ack=2485442141 Win=7920 Len=0
  6 2.984269 10.5.228.15 10.12.22.218 TCP 12049 > 8887 [ACK] Seq=2485442141 Ack=47020121 Win=0 Len=0
  7 6.664342 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020120 Ack=2485442141 Win=7920 Len=0
  8 14.025502 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020120 Ack=2485442141 Win=7920 Len=0
  9 28.747809 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020120 Ack=2485442141 Win=7920 Len=0
  10 58.192437 10.12.22.218 10.5.228.15 TCP 8887 > 12049 [ACK] Seq=47020120 Ack=2485442141 Win=7920 Len=0

-- 
dhoward@redhat.com
sustaining engineering


- To unsubscribe from this list: send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html



This archive was generated by hypermail 2b29 : Fri May 23 2003 - 22:00:02 EST