I'll post this just in case anyone is interested and, since the blade is
now back on it's legs after 9 long days, I'll blog about when I get the
chance. Here is what transpired:
The VIOS support guy I was working with couldn't get VIOS to boot,
despite a couple of days of fiddling with it. He finally suggested that
I order the latest VIOS install CD and reinstall VIOS to see if that
would fix it (ah! The tried and true Microsoft method: re-install). I
tried ESS and came up empty on an install image, then made about a half
dozen phone calls until a savvy entitlement person noted that I have
entitlements under both a "System i" and "System p" entry in ESS. Turns
out the System p downloads area was where I wanted to go. It was subtle
enough that I and two ESS folks didn't see it.
5 hours later I had the VIOS new install DVD image downloaded. I
started the installation at about 8pm and at about 10pm I noticed that
it stopped progressing at 86% at the "Copying Cu* to disk" stage. This
morning it was still at the same point so I figured it had hung and I
restarted the install.
A couple of hours later I notice that it had finished so I restarted
VIOS and logged in (new userid and password). I had to accept the VIOS
license agreements again and I had to configure the networking, then I
bounced the server and now I had the IVM back.
The i5/OS partition was still there but in checking the properties I
didn't have any hard disks configured for the partition any more. A
moment of panic ensued while I wracked my brain trying to remember just
what the heck I had originally configured. After poking around a bit I
remembered that all the RAID stuff is done on the i and I remembered
having two, mirrored pairs so I decided to cross my fingers and give it
4 of the SAS drives in the BCS. The application hung for about 5
minutes after I clicked OK but finally it displayed the partition info
and I clicked on the "Activate" button for the partition. Almost
immediately the BCS reported errors and phoned home, indicating a
hardware failure and no load source. I assumed I must have configured
the partition incorrectly so I attempted to open up the partition
properties and, after a long pause, the console started spilling errors
into the IVM. From that point on the blade was pretty much a doorstop.
So bounced the blade and decided to install the latest fix pack for VIOS
2.1. The command prompt took a long time to return after the bounce but
I was able to start and finish the update. After a restart and a long
boot cycle I got the IVM back and checked on the i5/OS partition
information. This time the configure disks screen displayed almost
instantly and I selected the 4 SAS disks and then clicked OK. It saved
the info and I double checked by clicking on the partition properties
again, just to make sure. Then I attempted to restart the i5/OS
partition. After about 10 minutes is was up and running.
However, I couldn't connect to it and again, after poking through the
blade readme and some other information I had, I guessed that the
virtual networking wasn't set up properly. I followed the readme
instructions again and assigned the virtual Ethernet Bridge to the
Ethernet line I had configured in VIOS and voila, the i was back!
So, it looks like if you re-install VIOS you need to reassign disks and
networking resources to the the i partition again. All my other
partition selections were retained. Don't know if this is working as
designed or if it is a feature.
But, at least it is back. Just in time for me to work on my Common
presentations......
Pete
Pete Helgren wrote:
I am just passing this on in case any of you early adopters on blades
end up with a similar situation:
My JS12 had a few intermittent episodes where it would just stop
responding. The blade itself would continue to run but VIOS and i5/OS
would no longer respond. The Blade Center itself has an error log and
errors would show up there. However, I never got a call from IBM
letting me know it had phoned home. Not big deal, bouncing the blade
seemed to take care if it.
Only, it didn't. The problem continued and last weekend after the blade
crashed again, I finally figured out that the i5/OS ECS line wasn't
working, and despite my best efforts, couldn't get it to configure and
phone home a test problem (that is another story). However, I got the
Blade Center configured to phone home and I lit up the blade again,
sending a test problem. All was well. Until last Monday morning when I
discovered the blade was DOA again. This time IBM did call but I was
working with the Blade Center group who seemed to think that there was a
part that needed replacing wholly unconcerned that the system was down.
The sent me the part and it was replaced, two days later.
That didn't take care of the problem and now the blade is really down
and these folks seem not to be as "responsive" and the "real" system i
folks have been in the past. They know that there is an issue with the
Media Tray causing a system to crash because other customers have
reported the same issue but they haven't come up with a patch/workaround
to the problem. This is a development machine so it isn't like the
business can't run, but I am a little surprised by the lack of urgency
in trying to quickly resolve the problem. There is of course much
correspondence and more to the story that I can give here, but this is a
bit disconcerting. Even on my development machines in the past IBM
would dispatch someone pretty quickly.
Anybody on a blade have a different (positive) experience? Have you
always gone to the i group first? This is definitely a
hardware/firmware issue but I think I should go through the IBM i group
next, rather than wait on these blade folks to figure it out.
Thoughts?
Pete Helgren
midrange.com
