SUMMARY: Setting up for remote installation

From: Patrick L. Nolan (pln@razzle.stanford.edu)
Date: Fri Apr 28 2000 - 11:48:57 CDT


There have been 11 answers so far. Many have focused on the issue of
NFS sharing the cdrom, which turns out to be the key. I didn't
explicitly mention it in my original message, but add_install_client
produces this entry in dfstab:
share -F nfs -o ro,anon=0 /cdrom/sol_7_sparc
After running "shareall", the cdrom should have been visible to
the client. It didn't work. Richard Cove was the first to point
out that it might help to unshare the disk and then share it again.
That fixed the problem. For some reason, the first share just didn't
work, and the system needed to be pounded a bit to wake it up.

<rant> This type of apparently non-deterministic behavior by NFS and
the automounter seem to be fairly common. I don't know how many hours
I have wasted on this type of thing.

Original question:

> I have several Sparc workstations that need to be upgraded to Solaris 7.
> They don't have CD players, so I have been studying the Answerbook to
> learn how to do a remote upgrade. It isn't working, so I must have
> misunderstood something.
>
> I decided to make it as simple as possible. Everything is on the same subnet,
> so I should be able to install directly from a CD on a server. I executed
> this command on the server:
> add_install_client myclient sun4m
> It added appropriate entries to /etc/bootparams and nsswitch.conf, and
> it put some files in /etc/tftpboot. tftp is enabled. rpc.bootparamd
> is running.
>
> Then I shut down the client and booted it with the command
> boot net - install
> It speaks to the server enough to discover its hostname, server name,
> and NIS domain name. Then it says,
> root directory: /cdrom/sol_7_sparc/s0/Solaris_2.7/Tools/Boot
> panic - boot: Could not mount filesystem.
> Program terminated
> That pathname exists. It does the same thing for a plain "boot net".
> The tftp daemon is run through tcp wrappers, so the log file shows that
> it is being activated.
>
> I'm probably missing something simple.

* Patrick L. Nolan *
* W. W. Hansen Experimental Physics Laboratory (HEPL) *
* Stanford University *



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:06 CDT