nfs boot

[Oliver Schmitz]

Hi everybody,

I try to let an Agilent V743 controlled mainframe boot from our new HP-UX 11.11 i TCOE WS. I did the export of the booteble device via sam: SAM/Networking and Communications/Bootable Devices/Fixed Adress devices booting from this server. I gave the HW-adress of the controler. Unfortunately when the mainframe is searching after this bootpath it doesn`t find it at all.

Did I forget to enable any bootserver, is SAM not enable to do this operation properly, do I need a patch or is anything other wrong thing obvious?

Thanks for some hints in advance,

Oliver

[Steven E. Protter · Accepted solution]

ps -ef | grep boot

Make sure the boot helper/enabler is running./

Make sure the intended filesystem is in /etc/exports

If not add it and then

exportfs -a

You should be able to enable it without a patch, its just another service in /etc/inetd.conf

Enable there and...

inetd -c

Be careful about messing with this on a production machine.

SEP

[Oliver Schmitz]

Dear Steven,

thank you for these hints. They help as I am now sure that my steps I already did to bring this nfs export have been the right ones.

Everything you mentioned is running and well setuped.

Unfortunately my mainframe finds the server as its bootpath (obviously the server tells anything that a system is laying there) but when the controler start loading it it aborts without any message. The controler just wents back to the main menu where I further choose the bootpath.

I tried around a very long time now and at the end I copied the old diskless cluster structure from the old cluster server to my new maschine. I put ther uxbootlf file as the bootfile in the sam screen (and I checked the correct appearance in the /etc/bootptab). If I understand it right the mainframe should now load its old system due to this bootfile!!??!!
But it does not but the behaviour described above appeared again!

Is that way worth trying or are there any reasons why it will fail at all?

Do you or anybody else have some more ideas what the problem could be?

One informational question: Is there a reason why hp dicontinued the diskless cluster services? In my opinion this was and is on the running systems one extraordinary HP-UX tool and it is very, very unlikely that it is off now! May be someone else have similar impressions or a good reason.

Thanks very much to everybody for some more hints,

Oliver

[Bill Hassell]

You aksed:

> One informational question: Is there a reason why hp dicontinued the diskless cluster services? In my opinion this was and is on the running systems one extraordinary HP-UX tool and it is very, very unlikely that it is off now! May be someone else have similar impressions or a good reason.

True diskless workstations have been obsolete for almost a decade, and yes, they were very useful when a knowledgeable sysadmin (and fast server and network) were available. Diskless NFS was a marketing flop. Customers asked for a routeable diskless protocol (HP's diskless protocol was limited to a subnet) such as NFS and HP (among others) created the diskless code but response was underwhelming. It seems that extremely cheap disks and NFS stability problems make NFS clusters less useful than originally thought. NFS clusters suffered the same fate as thin clients, a popular concept a few years ago but a commercial drip as far as sales.