[comp.protocols.tcp-ip] CAMBRIDGE-MB and RESTON-DCEC-MB playing "Gateway ping-pong"?

trn@warper.jhuapl.edu (Tony Nardo) (01/30/89)

Since Thursday (1/27), I've noticed that the our link from the MILNET to the
ARPANET has once again become rather fragile.  Quite frequently, I've had
network connections drop.  Invariably, after being disconnected, an attempt
to "ping" the target node will yield either

	36 bytes from 26.21.0.104: icmp_type = 3 (Dest unreachable)
    or	36 bytes from 26.1.0.49: icmp_type = 3 (Dest unreachable)

followed by eight longwords in hex format.

Since my target node is always the same, it would seem that the Butterflys at
CAMBRIDGE-MB and RESTON-DCEC-MB are still grabbing control of network packet
routing from each other.

Perhaps more infuriating, I've noticed that if I "ping" the offending gateway
I can the reconnect to the target system almost immediately.  "pinging" each
of these gateways in the background also seems to make my connections to other
sites survive a little better.  (I don't *like* doing this, but I also don't
like watching ftp choke after I've transferred >200 Kb of a 300 Kb file. :-(  )

Question: is there any information that a person who is *not* on a gateway
system (like myself) can gather which would help track or solve this problem?
-------				----------------			-------
ARPA,	trn@aplcomm.jhuapl.edu,		UUCP:	{backbone!}mimsy!aplcomm!trn
BITNET:	trn@warper.jhuapl.edu

	programming (v): the art of debugging a blank sheet of paper.