FYI - Looks like our hardware vendor has found the answer (I'll know for sure a
bit later on):
After spending a while researching and thinking about the OSA problem, here's
what I think it is.
For the OSE cards, when TCP/IP is started, it uses the information in the
TCP/IP profile to load the port with IP address, etc, that the port needs to
connect to the network. The default configuration that gets loaded onto the
OSE chpids during POR has entries for unitadd 00 and 01, which makes the port 0
connections work. For port 1 the definitions are unitadd 10 and 11 for FC40
and FC10. Those unitadds do not get set up right in the card's default
configuration.
Here's the fix :
In HCD, add a device definition for a device type OSAD at unitadd FE to each of
the OSE chpids, like the following:
CNTLUNIT CUNUMBR=FC00,PATH=((CSS(0),00)),UNIT=OSA
IODEVICE ADDRESS=(FC00,032),CUNUMBR=(FC00),UNIT=OSA
IODEVICE ADDRESS=FCFE,UNITADD=FE,CUNUMBR=FC00),UNIT=OSAD <--
CNTLUNIT CUNUMBR=FC20,PATH=((CSS(0),04)),UNIT=OSA
IODEVICE ADDRESS=(FC20,016),UNITADD=00,CUNUMBR=(FC20),UNIT=OSA
IODEVICE ADDRESS=(FC40,016),UNITADD=10,CUNUMBR=(FC20),UNIT=OSA
IODEVICE ADDRESS=FCFF,UNITADD=FE,CUNUMBR=FC20),UNIT=OSAD
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
