[Sunhelp] Problem with E220 "redundant" power supplies

[Tony Alberti]

How many disks were in the 220?

I have had a similar situation on an E250.
I asked one of the Sun FE's here in Chicago
to explain this.  I was told that the dual power
supplies in a 250 aren't truely redundant when
the machine is loaded. If you have two power
supplies and only 1 internal disk, there is enough
power in one power supply to handle the load,
so the redundancy works.   However when you
have all 6 HDs in a 250, that second power supply
uses about 75% of its output to drive the disk
backplane.   It is no longer redundant.
If the machine had truely redundant power
supplies, one could handle the load of the
system AND all 6 disks, so all you would
need for operation was 1, and the second
for redundancy.   Try to order an E250
from Sun configured for 6 internal disks
and 1 powersupply.  They will NOT
sell you that configuration.  If you
want 2 internal disks and 1 power supply,
they will sell that to you (assuming the mainboard
had the same loaded config of 2 400mHz CPUS and
2 gig memory in both configurations).
I suspect something similar probably occured in your 220.
Although there are less internal disks in a 220, perhaps the
power supplies are less as well.

Tony

----- Original Message -----
From: "Bill Bradford" <mrbill at mrbill.net>
To: <sunhelp at sunhelp.org>
Sent: Wednesday, June 21, 2000 2:41 AM
Subject: [Sunhelp] Problem with E220 "redundant" power supplies


> For those of you who dont know, E220Rs have two power cords each, for
> redundancy - they can run fine w/one.  Anyway, here's a description of
> my strange problem.
>
> We have two UPS units in our datacenter right now, call them "A" and "B".
> Systems are plugged into both units right now, but tonight, we had to
> move everything to "A" because "B" is being shut down (to add more
batteries)
> this coming Friday.
>
> "no problem" I thought, "since all the 220s have dual power cords, we wont
> even have to shut them down - just do the shell game with the power cords
> until everything is plugged into UPS-A".
>
> It didnt quite work this way.
>
> Here's what happened.  Say each 220 has a power cord #1 and #2.
>
> 1.  All machines have power cords plugged into "first available outlet".
>
> 2.  I plug power-strip-1 into a UPS-A outlet.  One by one, I unplug
> Power Cord #1 from each 220 into power-strip-1.
>
> 3.  I unplug each 200s Power Cord #2 and let them sit for a minute while
> I untangle everything.  All machines are still up at this
> point, running off the power strip plugged into UPS-A.
>
> 6.  I take all the 220s Power Cord #2s, and plug them into
> power-strip-2.
>
> 7.  I unplug power-strip-2 from UPS-B.  All the E220s die, even tho
> they had been (and still should have been) getting power
> from power-strip-1, which was already plugged into UPS-A.
>
> 8.  I plug power-strip-2 into UPS-A.  All machines fire back up,
> and appear to come up normally. (whew)
>
> Therefore, it seems that the machines didnt failover between redundant
> power supplies/cords properly; if power-strip-1 was bad, they should
> have all died when I unplugged the power cord #2s - but they didnt.
> They died after I had all the power cord #1s going to UPS-A, all
> power cord #2s going to UPS-B, and unplugged power-strip-2 to move
> it to UPS-A.
>
> Thoughts?  Suggestions?  Anybody else seen this problem?  I cant
> take the boxes back down to "try it again", but I'm getting another
> E220 in a couple of days that I'm gonna try this with and see if it
> does the same thing.
>
> Bill
>
> --
> +-------------------\ /-----------------+
> | Bill Bradford      |  www.sunhelp.org |
> | mrbill at mrbill.net  |   www.decvax.org |
> | Austin, Texas USA  |    www.pdp11.org |
> +-------------------/ \-----------------+
> _______________________________________________
> SunHELP maillist  -  SunHELP at sunhelp.org
> http://www.sunhelp.org/mailman/listinfo/sunhelp