[1856] in Hotline Meeting
Problem with new printserver m24-618-p
daemon@ATHENA.MIT.EDU (sethf@ATHENA.MIT.EDU)
Thu Sep 27 01:45:30 1990
From: sethf@ATHENA.MIT.EDU
Date: Thu, 27 Sep 90 01:45:02 -0400
To: hotline@ATHENA.MIT.EDU
Cc: defisard@ATHENA.MIT.EDU
There is a problem with printing to the newly installed
printserver in 24-618. Printing works from three machines in the room,
but not two others. It is a very puzzling problem without much pattern
in its symptoms. The following is a log of an attempt to solve it over
olc.
============================================================================
Log Initiated for user Seth Finkelstein (sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0]).
[Mon 24-Sep-90 10:23pm]
Topic: workstations
Question:
What causes a line like this:
From /etc/arp -a
M24-618-P.MIT.EDU (18.62.0.50) at (incomplete)
Machine info: RTPC-ROMPC, apa16, 0x649800 user, 0x800000 (8 M) total
___________________________________________________________
--- Question grabbed by consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 10:42pm]
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 10:43pm]
That means the workstation has sent out an ARP request and is waiting
for a reply. Most likely the machine is down.
--- User sethf read reply.
[Mon 24-Sep-90 10:44pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 10:45pm]
No, it's fine. It's responding to requests from other machines in the
room.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 10:54pm]
Try arp -d 18.72.0.50
--- User sethf read reply.
[Mon 24-Sep-90 10:54pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 10:57pm]
No change. Same result.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 10:59pm]
I was just able to finger @m24-618-p@FRUMIOUS-BANDERSNATCH. Is it a
different workstation?
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 10:59pm]
You probably meant arp -d 18.62.0.50
I tried that also. Same effect.
--- User sethf read reply.
[Mon 24-Sep-90 11:00pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:02pm]
m24-618-p is reachable from 18.62.0.5{1,2,3}, but not 5{4,5}.
FRUMIOUS-BANDERSNATCH is 51.
I'm very puzzled by this. I thought at first it was an Ethernet artifact due to
m24-618-p being hooked up between 53 and 54, but moving the connection to the
end of 55 didn't fix things.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:03pm]
I mistyped "72" for "62" in the arp command.
--- User sethf read reply.
[Mon 24-Sep-90 11:03pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:03pm]
Yes, I noted that and tried it with 62 - no effect.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:23pm]
Have you ever used netwatch?
--- User sethf read reply.
[Mon 24-Sep-90 11:24pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:24pm]
A few times.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:26pm]
It may be a useful tool here. If packets are only being lost in
one direction, netwatch can detect this. (Run netwatch on one of
the workstations that can't talk to the server, and arrange for the
server to send a packet [for example, indirect finger]. If you see
a packet or ARP request, then the lossage is only one way).
--- User sethf read reply.
[Mon 24-Sep-90 11:26pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:27pm]
Where is netwatch?
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:27pm]
nit
--- User sethf read reply.
[Mon 24-Sep-90 11:27pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:33pm]
I have netwatch running, past this I need guidance. I see streams of
<number> PACKETS LOST **
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:35pm]
Type "m arp" to watch only ARP requests and replies.
--- User sethf read reply.
[Mon 24-Sep-90 11:36pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:41pm]
done, but I'm having a hard time using the output.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:42pm]
Do you see anything coming from the 18.62.0.50?
--- User sethf read reply.
[Mon 24-Sep-90 11:42pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:44pm]
Yes.
Req from 50 to 55,
Reply from 55 to 50.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:44pm]
Are they talking normally now? Does arp -a still say "incomplete"?
--- User sethf read reply.
[Mon 24-Sep-90 11:45pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:47pm]
Deleted arp.
Tried to talk from 55.
ARP Req from 55 to 50, no reply.
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:49pm]
Doesn't say (incomplete), but still not talking.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:50pm]
That suggests that packets are being lost only in one direction -- from
.55 to .50. I'm not sure what can be done about this,
--- User sethf read reply.
[Mon 24-Sep-90 11:50pm]
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Mon 24-Sep-90 11:50pm]
All these machine are connected to the same box?
--- User sethf read reply.
[Mon 24-Sep-90 11:50pm]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Mon 24-Sep-90 11:51pm]
Same box, same room, same chain. It's very, very odd.
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Tue 25-Sep-90 12:13am]
But why is it only a .5{4,5} to .50 problem? And why are there no other
connection problems? It doesn't make sense.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Tue 25-Sep-90 12:14am]
I don't know. If there were more hardware between the machines I
could understand it (we've had problems with repeaters getting
confused).
--- User sethf read reply.
[Tue 25-Sep-90 12:15am]
*** Reply from user sethf@FRUMIOUS-BANDERSNATCH.MIT.EDU [0].
[Tue 25-Sep-90 12:30am]
I thought it might be the cables. I tried taking off .55 and
terminating the net at .54 . Still didn't work. Checking the records, I see
.54 and .55 have never successfully communicated with .50 . I'm utterly stumped.
*** Reply from consultant jfc@ACHATES.MIT.EDU [0].
[Tue 25-Sep-90 11:42am]
You should mail a report to hotline (you may want to include this log).
