I have a E420R with 4x 450Mhz CPUs. In an effort to conserve power and
air conditioning in my computer room, I decided to yank two of the
CPUs.

According to the manual, running a E420R with two CPUs is fine, but you
have to remove CPUs 0 and 3, which I did, leaving the two middle CPU
modules. Upon rebooting the server, I get a blank white screen. No logo
in the upper left. Nothing. I thought maybe there was some kind of
jumpers or something I might have missed, but there is no mention of
anything of the sort in the manual. It does recommend putting blank
placeholders in the empty CPU slots and I don't have any of these,
however it's hard for me to imagine that they do anything.

Any thoughts?

What's worse is that putting the two removed CPUs back and restarting
results in OpenBoot starting up normally, and then freezing after the
memory scan. This computer worked fine prior to this!!

Bruce C. Miller wrote:
> I have a E420R with 4x 450Mhz CPUs. In an effort to conserve power and
> air conditioning in my computer room, I decided to yank two of the
> CPUs.
>
> According to the manual, running a E420R with two CPUs is fine, but you
> have to remove CPUs 0 and 3, which I did, leaving the two middle CPU
> modules. Upon rebooting the server, I get a blank white screen. No logo
> in the upper left. Nothing. I thought maybe there was some kind of
> jumpers or something I might have missed, but there is no mention of
> anything of the sort in the manual. It does recommend putting blank
> placeholders in the empty CPU slots and I don't have any of these,
> however it's hard for me to imagine that they do anything.
>
> Any thoughts?
>
> What's worse is that putting the two removed CPUs back and restarting
> results in OpenBoot starting up normally, and then freezing after the
> memory scan. This computer worked fine prior to this!!

I'll also add that with a single CPU (in slot 1) it boots up to the
"Initializing Memory" stage and then hangs as well. I tried setenving
the diag-switch to true and then watching it carefully after resetting,
and everything seems normal. The memory counts down to 0 and then the
system freezes and is unresponsive to input.

Bruce C. Miller <bm3719@gmail.com> wrote:
> I have a E420R with 4x 450Mhz CPUs. In an effort to conserve power and
> air conditioning in my computer room, I decided to yank two of the
> CPUs.

> According to the manual, running a E420R with two CPUs is fine, but you
> have to remove CPUs 0 and 3, which I did, leaving the two middle CPU
> modules. Upon rebooting the server, I get a blank white screen. No logo
> in the upper left. Nothing. I thought maybe there was some kind of
> jumpers or something I might have missed, but there is no mention of
> anything of the sort in the manual. It does recommend putting blank
> placeholders in the empty CPU slots and I don't have any of these,
> however it's hard for me to imagine that they do anything.

> Any thoughts?

> What's worse is that putting the two removed CPUs back and restarting
> results in OpenBoot starting up normally, and then freezing after the
> memory scan. This computer worked fine prior to this!!

Yeah, you disloged the memory or the backplane enough to cause a
problem. Have you checked both of those? I dealt with a lot of 420Rs
and sometimes had problems with this just when people moved them from
rack to rack. The connecting system for that backplane doesn't seem to
be the most reliable design and if I remember right it was also
connected at least partly to the CPU cage as well, you could have
disrupted it just by pulling the CPUs. There are some screws holding
the whole CPU cage down, make sure those are tight, also the two wacky
screws on the memory backplane. You're supposed to use a torque driver
on these according to Sun and there might even be one stashed inside the
machine, but if not careful hand tightening is probably ok.

I don't know why they actually split each bank across that
backplane instead of making two banks on the main board and two on the
back plane. If they had done that, at least you could have ripped out
the backplane or not even used it if you didn't need the extra RAM.

Bruce C. Miller wrote:
<snip>
> It does recommend putting blank
> placeholders in the empty CPU slots and I don't have any of these,
> however it's hard for me to imagine that they do anything.
>
The blank placeholders things are to should be inserted to ensure
correct airflow so they do do something.
--
Chris

Bruce C. Miller wrote:
> I have a E420R with 4x 450Mhz CPUs. In an effort to conserve power and
> air conditioning in my computer room, I decided to yank two of the CPUs.
> According to the manual, running a E420R with two CPUs is fine, but you
> have to remove CPUs 0 and 3, which I did, leaving the two middle CPU
> modules. Upon rebooting the server, I get a blank white screen. No logo
> in the upper left. Nothing. I thought maybe there was some kind of
> jumpers or something I might have missed, but there is no mention of
> anything of the sort in the manual. It does recommend putting blank
> placeholders in the empty CPU slots and I don't have any of these,
> however it's hard for me to imagine that they do anything.
>
> Any thoughts?

Yes I too have been surprised by the E420r in the past.
As someone already mentioned the empty slots require a "blank"
CPU placeholder for correct airflow. Incorrect airflow would likely
cause MORE heat generation defeating your rather dubious original
intent!
Disassembly/assembly of the 420r requires a fair amount of caution
and CPUs need a special torque wrench for re-installation.
In short these boxes should not be monkeyed with : <
If you havent cracked any traces or something, careful reassembly
should get you back on line. Until then think of all that power you
have saved.. : /

gerryt@ wrote:

> Yes I too have been surprised by the E420r in the past.
> As someone already mentioned the empty slots require a "blank"
> CPU placeholder for correct airflow. Incorrect airflow would likely
> cause MORE heat generation defeating your rather dubious original
> intent!
> Disassembly/assembly of the 420r requires a fair amount of caution
> and CPUs need a special torque wrench for re-installation.

If its like the Ultra 80 (which I think it is), the torque wrench is
needed for memory, not CPUs.

--
Dave K MCSE.

MCSE = Minefield Consultant and Solitaire Expert.

Please note my email address changes periodically to avoid spam.
It is always of the form: month-year@domain. Hitting reply will work
for a couple of months only. Later set it manually.

> If its like the Ultra 80 (which I think it is), the torque wrench is
> needed for memory, not CPUs.

The E420R is basically the server version of an Ultra 80. Same system board,
memory, CPUs, etc...

The torque wrench is for the memory so if only CPUs were removed and it was
working fine before, then I doubt it's memory.
Now the Sun Fire 280R uses the same style torque wrench for the CPUs so
perhaps someone confused the two?

Trinean

Trinean wrote:
> > If its like the Ultra 80 (which I think it is), the torque wrench is
> > needed for memory, not CPUs.
> The E420R is basically the server version of an Ultra 80. Same system board,
> memory, CPUs, etc...
> The torque wrench is for the memory so if only CPUs were removed and it was
> working fine before, then I doubt it's memory.
> Now the Sun Fire 280R uses the same style torque wrench for the CPUs so
> perhaps someone confused the two?

You may have me there. It was a couple of years ago when I was
confronted
with a stack of E420r's and tasked to "evaluate" them. Some parts were
exchanged
to beef up a couple of boxes intended for some project and some part
swaps
resulted in 1 dead machine as I recall.
Perhaps it was already dead!
These part swaps were both memory and CPUs. There were no
torque wrenches about... Point is these machines seem to be delicate
and
require more than the usual care : / Sorry for the FUD!


>
> Trinean

B. Wright wrote:
> Yeah, you disloged the memory or the backplane enough to cause a
> problem. Have you checked both of those? I dealt with a lot of 420Rs
> and sometimes had problems with this just when people moved them from
> rack to rack. The connecting system for that backplane doesn't seem to
> be the most reliable design and if I remember right it was also
> connected at least partly to the CPU cage as well, you could have
> disrupted it just by pulling the CPUs. There are some screws holding
> the whole CPU cage down, make sure those are tight, also the two wacky
> screws on the memory backplane. You're supposed to use a torque driver
> on these according to Sun and there might even be one stashed inside the
> machine, but if not careful hand tightening is probably ok.

This is probably something like what happened. Running on a single CPU,
I pulled riser board and retightened all the CPU bracket screws then
replaced it to no effect. I then pulled half the memory and it booted
up ok. Probably a loose connection on one of the memory sticks I guess.
Unless one of them went bad.

> I don't know why they actually split each bank across that
> backplane instead of making two banks on the main board and two on the
> back plane. If they had done that, at least you could have ripped out
> the backplane or not even used it if you didn't need the extra RAM.

Yes, I noticed this as well. Rather dumb design, I'd say... But then,
there are plenty of examples of that in the e420r.