Newsgroups: comp.parallel.pvm
From: tim@osiris.usi.utah.edu (Timothy Burns)
Subject: FIN|ACK = Help! Ack!??
Organization: Utah Supercomputing Institute, University of Utah
Date: Fri, 3 Jun 1994 20:33:36 GMT
Message-ID: <TIM.94Jun3133336@osiris.usi.utah.edu>


Hi.

I am wondering if anyone out there has encountered the following:

	[t80040000] netinput() FIN|ACK from graph
	[t80040000]  hd_dump() ref 1 t80000 n "graph" ar "RS6K" lo ""
	[t80040000]            sa 128.110.138.162:2905 mtu 4096 f 0x0 e 0 txq 0
	[t80040000]            tx 40990 rx 1 rtt 0.006233
	[t80040000] netinput() bogus pkt from 128.110.136.13:53
	[t80040000] netinput() bogus pkt from 128.110.136.13:53
	[t80040000] netinput() bogus pkt from 128.110.136.13:53
	[t80040000] netinput() bogus pkt from 128.110.136.13:53

The bogus packet we are seeing, comes not from graph, but from
cornice.usi.utah.edu, which is a nameserver.  I have seen the
latter before, and ignored it, but the former, (NET|ACK) I cannot
ignore, because it corresponds to the disappearence of my process.

Here is some background:

~25000 messages were passed, the program ran for 00:32:08:927 
xab3 timestamps, and each message size was 16.8 kb - 27.72 kb.

This same program runs fine for fewer timesteps (there are ~35000 time
steps), my checkpoint files somehow didn't dump, so I don't know the
timestep.  It looked like it got about 1/4 of the way.  The same
program also runs fine for smaller models (translate message sizes).

Anyways, we have seen this problem for models larger than 4MB per 
processor, and I am wondering if others have experienced similiar
problems.

Thanks a lot.
Tim
--
Tim Burns				email: tim@osiris.usi.utah.edu
USI, 85 SSB, Univ. of Utah, UT  84112  	phone: (801)581-5172
+--------------------------------------------------------------------------+
| Even the most brilliant scientific discoveries will in time change and   |
| perhaps grow obsolete, as new scientific manifestations emerge.  But Art |
| is eternal; for it reveals the inner landscape which is the soul of man. |
+----------------------------------            --Martha Graham    ---------+


