pung@ornl-ncc.ARPA.UUCP (01/30/87)
We have a Delni which is configured as follows:
1. Several directly connected VAXes
2. An Interlink DEC/IBM gateway (Qbus Micro PDP-11)
3. An Applitek Bridge which supports an extended LAN
across a three plant network. Note that the
Applitek Bridge is connected into one of the normal
ports on the Delni not the "control" port.
Everything was working fine until we connected another segment
of Ethernet into the "control" port of the Delni.
With the Delni switch in the "Global" position, the Micro PDP-11
console now continually displays NCP error messages along the
vein of "Collision Detect Test Failed" and "Short circuit".
However when the Delni switch is in the "Local" position, the
Micro PDP-11 is extremely happy and quiet.
After much discussion and testing, we have theorized the following:
The Deqna on the Micro PDP-11 is sending out a heartbeat
pulse and depending on the position of the switch on
the Delni the heartbeat is either responded to (turned around)
by the Delni ("Local" setting) or the heartbeat is not
answered by the Delni and/or the rest of the network ("Global").
Apparently there is no way to disable the heartbeat function on
the Deqna and since it is connected to the network via a Delni
instead of an H4000, no one ever responds to the heartbeat and
error messages result.
My questions are the following:
1. Are our theories right or are we off base?
2. Is there a way to turn off the heartbeat function of a Deqna ?
3. Has anyone else seen this problem, (i.e. Deqna to Delni in "global")
4. Would changing some pins in the transceiver cable possibly disable
the sending of the heartbeat signal into the Delni ?
Any help, comments, suggestions would be greatly appreciated.
Mike Pung
Oak Ridge National Lab
Pung%ornl-ncc@ornl-msr.arpa
------JMS@ARIZMIS.BITNET.UUCP (02/01/87)
Recently Mr. Pung (PUNG@ORNL-NCC.ARPA) wrote about some problems he's having with heartbeats and DELNIs. I don't have a general-case solution, but we are working on the exact same sort of problem here at Arizona, in this case the connection of Ethernets using American Photonics bridges, some American Photonics transceivers, some H4000s, and DEBETs (Lan Bridge 100s). The symptom I observe is a half-duplex connection between adjacent ethernets, with a full duplex connection between non-adjacent Ethernets. I'm literally in the middle of debugging the thing, but one solution seems to be matching transceivers with equipment from the same manufacturer, since all Heartbeat claims appear to be untrustworthy. Hope this helps, jms (Joel M Snyder, University of Arizona MIS, JMS@ARIZMIS.BITNET)