Without knowing the OS and if you have path redundancy configured correctly (i.e. running PowerPath or using native OS path redundancy) can't give a complete answer.

However - assuming you have PowerPath or an equivalent up and running properly than the answer should be that the host should have no issue with an HBA being disconnected.

The only 2 things I can think to consider:

1) You will have some error messages generated for the lost path. Just watch for any monitoring software (Tivoli, Openview, etc.) to generate some potential "false alerts" if you have an operations staff.

2) If you have a VERY heavily I/O loaded server than dropping a path could have some performance impact to your host as you effectively half your potential bandwidth.

OS is solaris 5.9 and the multipath software is powerpath Version 5.0.1 (build 25).
i didnt get the point that " performance impact to your host as you effectively half your potential bandwidth". Node wwn# zoning means the whole HBA is zoned to FA instead of individual port in it. thus, i/o will be distributed across the ports in it, hence the performance will be better. not sure though, it was my assumption. pls clarify. thanks

it is also noticed that both port wwn# and node wwn# were added to the initiator alias and zoned to FA.

But only the port wwn# is logged on in fabric and array, not the node wwn#.
symmask -sid xxx list logins

You only need the wwpn, not the wwnn. As you already noticed the node wwn doesn't do a FLOGI anyway !

what is the HBA model? is it is dual/multi port card then you will have multiple wwpn ,one for each port

I may have misunderstood your initial question - I read it as you had multiple HBAs connected and zoned into multiple FA ports which is where my comment on losing 50% of your bandwidth came from.

As RRR suggested you don't need to zone the node wwn.

Do you have multiple HBA ports involved in this scenario or just the one card/port using the port and node wwn on the card?

just the one card/port using the port and node wwn on the card. as i have highlithed already, both wwnn and wwpn were zoned FA port. i can clear it later if no impact due to this kind of zoning. thanks

it depends, what do you have in your sd.conf and lpfc/qla.conf ? The reason i ask i had a similar issue where HPUX host was zoned to an STK array. It was zoned using its node WWN instead of port WWN, the host saw 3 hardware paths instead of just the 2 paths. I would be very careful to see what paths will change on the host once you rezone it correctly, i don't think Solaris creates any paths at will if you are using persistent binding ..but something to check.