On 10/3/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
Should we do the primary server first, along with two of the secondary servers,
reboot and do the rest of the servers after testing the patch? Or should we
patch each server one at a time then start the firmware updates after all have
been rebooted?
I would patch one of the secondaries first. Mark it "offline" ('utadm
-f') as far
ahead of the planned downtime as possible so that user sessions naturally
migrate away from it, leaving (hopefully) relatively few sessions to be
kicked off when the time comes. When you're ready to apply the patch turn
off firmware downloads on all of the servers (just like the patch README
says) then apply the patch, reboot and leave the secondary "offline" until
you've tested it and satisified yourself that the patch is solid. (Test it by
utswitch'ing or utselect'ing to this server. The "offline" status should
prevent sessions from being placed there automatically but an explicit
utswitch will place a session there.)
When you're satisfied that the patch is good then mark this server "online"
('utadm -n') and roll through the rest of the secondaries, marking each one
"offline" as soon as you've finished the previous one. Do one at a time over
a period of a few days or weeks. If you let people know what the schedule
for the remaining secondaries is they can plan their migrations around the
schedule, so nobody should have to log out more than twice and nobody
should have to get kicked off at an inconvenient time. When you've done
all of the secondaries you can re-enable firmware downloads (there's no
need to wait for the primary if it doesn't participate in offering firmware)
and then finish up by patching the primary. Try to do that at a quiet time,
because while the primary is down no changes can be made to the Data
Store contents.
Testing the patch firmware is a little tricky. The easiest way is to set up a
private interconnect subnet off a patched standalone test server or off the
first patched secondary and configure firmware downloads for that subnet.
Then you can put the patch firmware into selected Sun Rays by plugging
them into that subnet, then carry them back to the live subnet. If you
can't set aside a subnet like that then you can configure DHCP on the
patched server to deliver firmware updates only to selected DTUs. That's
a more complex process.
OttoM.
__
ottomeister
Disclaimer: These are my opinions. I do not speak for my employer.
_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users
