2 DEGPA on ES40 switching DECnet

[Antoniov.]

Jeff,
it's weird on node A you have physycal address different from hardware address meanwhile on node B they are the same.
Are you sure you are running DecNet IV on both nodes?
It seems on nodeB it's running DecNet phase V (or DecNet over TCP/IP).

Antonio Vigliotti

[Jeff Lanka]

phase 4 on both nodes

[Antoniov.]

Jeff,
I guess installation on node A has made in different way from node B.
If I remember I've meet similar problem on DS10 with 3 NIC card.
I sugget you remove ewab on node, then reconfigure decnet using netconfig_update.
Here command example:
NCP>PURGE LINE EWA-1 ALL
$ @SYS$UPDATE:NETCONFIG_UPDATE
Obviuosly you cannot execute thees command while user are connected via decnet.

If you have some confidence with NCP you could proceed full manually:
NCP>PURGE LINE EWA-1 ALL
NCP>DEF LINE EWA-1
Number of receive buffers (1-1024):
Protocol (DDCMP CONTROL/DMC/POINT/TRIBUTARY, LAPB, LAPBE, ETHERNET, FDDI, TOKEN
RING):
Line state (ON, OFF, SERVICE): OFF
NCP>SET KN LINE STATE OFF
NCP>SET KN CIRC STATE OFF
NCP>SET EXEC STATE SHUT
$ @SYS$MANAGER:STARTNET

Antonio Vigliotti

[Volker Halle]

So the answer is...

If MOP is disabled (NCP SET CIR EWA-0 SERVICE DISABLED), there is no response from the other DEGPA (EWA-0) using the SAME address (i.e. the DAT test does not see a response) and you can enable the 2nd device (EWA-1) using the SAME physical address with NCP SET LINE EWA-1 STATE ON. You can even turn on the circuit EWA-1 and turn off circuit EWA-0 and DECnet continues to work.

Whether this is supported on a DECnet Phase IV end node may be another question. As soon as you would start other protocols on that DEGPA, you might get into trouble. You now have effectively 2 NICs with the SAME address on the SAME LAN segment !

The 'disabled by software' blinking light seems to reflect the RUN or INITED device state. Once you stop the last active protocol on the DEGPA, the driver turns off the RUN bit and the green light starts blinking.

Volker.

PS: I've tested this in the running system. Now I can't seem to get the NCP SET LINE EWA-n to fail anymore with IVADDR. Even if both DEGPA circuits are SERVICE ENABLED.

[Jeff Lanka]

I found that once I forced one of the machines that were giving me the IVADDR errors into switching to the other DEGPA (by disconnecting the cable) it would then switch back and forth happily with no more errors.

[Volker Halle]

UPDATE:

MOP does not seem to have anything to do with this. Further testing has shown:

If you turn off all protocols on the LAN device, the RUN bit gets cleared and the physical address used 'falls back' to the Hardware Address of the NIC.

If you turn on the DECnet Phase IV LINE in such a state, it does NOT fail with IVADDR and starts using the DECnet address AA-00-04-00-nn-nn as the physical address. The 'DAT xmt complete' time shows that a DAT test has been done, but the 'DAT rcv found' will not show an updated timestamp (check with SDA> SHOW LAN/DEV=Exc on 4th page).

Volker.