Messages in this thread | | | Date | Thu, 19 Mar 1998 10:25:25 +1100 | From | Richard Gooch <> | Subject | Re: TCP stalls in 2.0.33 and 2.1.x |
| |
David S. Miller writes: > Date: Thu, 19 Mar 1998 00:16:22 +1100 > From: Richard Gooch <rgooch@atnf.CSIRO.AU> > > The MTU on my SLIP link is 255. I'll check tomorrow what the CISCI > terminal server is set to, but I think it should be higher than > that. Would that cause a problem? > > SunOS Advertised an MSS of 1460 (ethernet frame size) this must mean > it belives the Linux box is on the local subnet and also connected to > the same ethernet segment. SunOS doesn't implement path-MTU > discovery. So it sent the full sized frames. > > So the conclusion I'm at right now is that either: > 1) the CISCO is dropping the packets completely, or... > 2) the CISCO is not fragmenting them correctly, or... > 3) the Linux box is not un-fragmenting them correctly, or > just plain not receiving them at all > > And as usual, this could be yet another VJ compression bug of some > sort at either the CISCO or the Linux box.
I've been told that the MTU for the CISCO was 1500. I think that's what I had it configured to when we got it recently, but then I lost that in a recent disc crash and then I set my MTU to 255 (what is was for our old Annex). This would also explain why I've only noticed this problem recently.
Anyway, I've set machine A to have MTU=1500 for the SLIP link, and these TCP stalls seem to have gone away.
Regards,
Richard....
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |