Summary: using disksuite to clone a system

Thanks to all who responded

As it turns out, since both systems are identical hardware this cloning
process is reasonably simple.  

The following summarizes the procedure we have successfully developed and
tested for cloning, based on the advice I received.

initial assumptions:

system 1 is the original Solaris 5.8 unit, system 2 is the new clone we are
making.

each server has 2x18 Gb drives, each with 3 partitions.  

All drives are identical model#s.

There are 3 metadatabases on each drive slice 7

The md.tab looks like this:

d0 -m d1 d2 1
d1 1 1 c0t0d0s0
d2 1 1 c0t1d0s0
d30 -m d31 d32 1
d31 1 1 c0t0d0s3
d32 1 1 c0t1d0s3
d50 -m d51 d52 1
d51 1 1 c0t0d0s5
d52 1 1 c0t1d0s5


Procedure on system 1

1. metadetach d0 d2; metadetach d30 d32; metadetach d50 d52
2. metaclear d2; metaclear d32; metaclear d52
3. remove second drive (c0t1) from system while hot. This will become the
new root drive for system 2
4. metadb -d -f c0t1d0s0
4. put new drive in second slot (c0t1) and wait for it to spin up and be
visible to OS
5. format with "prtvtoc -h /dev/rdsk/c0t0d0s2 | fmthard -s -
/dev/rdsk/c0t1d0s2"
6. metadb -a -c 3 c0t1d0s0
7. metainit d2 1 1 c0t1d0s0; metainit d32 1 1 c0t1d0s0; metainit 52 1 1
c0t1d0s0
8. metattach d0 d2; metattach d30 d32; metattach d50 d52
9. after mirrors resync, reboot.  The original system is identical to how we
started now.

procedure on system 2

1. while powered off remove any existing drives and install the drive
removed from system 1 into the first drive slot (c0t0) on this system.
2. power up and allow to autoboot.  It will fail with metadb errors throwing
you into the miniroot shell.
3. metadb -d -f c0t1d0s0
4. reboot -- -s (to get into normal single-user mode)
5. install another disk into second (c0t1) location
6. format it with "prtvtoc -h /dev/rdsk/c0t0d0s2 | fmthard -s -
/dev/rdsk/c0t1d0s2"
7. metadb -a -c 3 c0t1d0s0
8. metainit d2 1 1 c0t1d0s0; metainit d32 1 1 c0t1d0s0; metainit 52 1 1
c0t1d0s0
9. metattach d0 d2; metattach d30 d32; metattach d50 d52
10. after mirrors resync, reboot with init 6.  This system is now a clone of
system 1

additional notes:

This worked for my two tests using a Netra T1 first and then an 1120.  In
the real case I will be on Netra 20's so the drives will be 36Gb.  Also note
the Netra20 uses C1 rather than C0.

----------------------------------------------------------------------


Original question:
to the esteemed wizards...

I have a project that requires upgrading all of the VOIP application
software on a 5.8 Netra 20-server that is in 24x7 service with small
maintenance window opportunities.    Management is requesting we use a
duplicate system to do the upgrade, leaving the original fully intact in
case of problems.   The condundrum is that we do not have time to clone the
box onto a second and do the upgrade all within the prescribed maintenance
window.

Because the disks in this box are all mirrored with DiskSuite, it has been
suggested to simply take out one of the drives for each mirror pair and put
them in the duplicate system.  Then we would do the upgrade on the second
system and put new spare drives in both to re-mirror them after the
application upgrade is completed. 

On the surface this sounds simple however my prior experiences with
disksuite tell me this is risky. I'm sure it is easy to corrupt the
partitions to a nonfunctional state.   Obviously I may need to add or delete
metadatabases but what other issues might occur?

Has anyone done this and/or has anyone written a procedure that I can
follow?

Thanks 
Jon
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers