[mod.computers.vax] Merging VAXen

blarsen@OSLO-VAX.ARPA (Bj|rn Larsen) (01/21/86)

I am managing a mini-cluster of two 11/750s.
Up to now, we have been managing this system as a heterogenous cluster,
with separate SYSUAFs for the two machines.

Now we want to make the system homogenous. This involves merging
the two SYSUAF files to one file, placing the resultant SYSUAF
in SYS$COMMON:[SYSEXE].

Does anybody have any hints on how to perform this in such a way that
the users retain their profile, complete with passwords?

(All user-profile settings should remain the same, including
logged-in directory, username, quotas, privileges etc.)

-----------------------------------------------------
blarsen@oslo-vax.ARPA             Bj|rn Larsen
blarsen@norunit.BITNET            System Manager
Bjorn_Larsen_UiO@qzcom.MAILNET    Computing Centre
                                  University of Oslo, Norway

(Computers never make misteaks ;-)

McGuire_Ed@GRINNELL.MAILNET (01/27/86)

Follow the procedure in the Guide to VAXclusters, appendix A.  If I remember
right, the same procedure can be used on your VMSMAIL.DAT file.  You will also
need to initialize your queue file (JBCSYSQUE.DAT).  You may want a common
SYSTARTUP.COM and SYLOGIN.COM since the systems are going to be largely the
same.  Differences between systems (such as different hardware) can be coded
in the same file if you look up the system's node name.  My procedures contain
code like this:

        $ AVAX = "!"
        $ BVAX = "!"
        $ CVAX = "!"
        $ 'F$GETSYI("NODENAME")' = ""
            .
            .
            .
        $ 'AVAX' SET PRINTER ...   ! SET PRINTER ONLY IF STARTING NODE AVAX

Merging your SYSUAF and VMSMAIL files is tricky if you have users on both nodes
nodes with the same name.  Only the first record encountered during the merge
is entered in the common file; therefore (for SYSUAF) only one profile is saved
and (for VMSMAIL) the new mail count can be wrong and only one set of users'
personality settings are saved.

Disclaimer:  The above information was written from user memory without ECC.
I guarantee that I have made mistakes and forgotten things.  Call CSC and
review the merge with the cluster support group before you start.  (You ARE on
maintenance, right?)  This is something they're good at.