Need to test server PSU but unsure how

[auphil]

Hi Guys,

I have a server from the now defunct Monarch Computers. This is a system with a Tyan S2882 motherboard and a Chenbro RM312 1+1 chassis.

It was cranking along happily until Monday of this week when I got a notice that the site it serves is down. I went into the data center and saw that the system was powered off. I felt lucky and just powered it back on and it booted normally. I thought, at the time, that one of the sudoer's had mistakenly sent a shutdown command instead of a reboot command.

Today, I got another notice the site was down. Now when I try to power it on, it is dead - nothing. Will not power on. I checked up (APC 3000 SmartUPS), it's fine.

This unit has triple redundant PSUs so I pulled all three plugs, then plugged them back in. When plugging in the first one, I head a single high pitched beep. No beeps on plugging in PSUs 2 and 3.

After replugging in, still no luck on powering up. I'm thinking one of the 3 PSUs (or more) is dead. I have an ATX PSU tester but these are obviously slide in hot-swappable PSUs so I'm not sure how to test them.

The model number is EMACS GIN-6350P

Does anyone know of a PSU tester for these guys?

Thanks,
Phil

[ChrisH]

Sounds like you are going to need a multimeter and figure out which pins to short to turn them on.

[kestrel1]

Normally if you have redundant PSU's only one is doing any real work. The others are there to take over if one goes down. This system may be different of course. It sounds more like the mobo or another component that has failed.

[auphil]

Well, after further testing (pulled from the rack and on the bench now), the system starts a boot (fans spin up, lights come on) when I have only ONE PSU in. If I put all 3 in, then apply power, it's still dead.

I've tried rotating the 3 PSUs in various slots and each one gives about 3-5 seconds of system power (again, fans spinning and lights on) before dying.

I've got 2 replacements coming in on overnight shipment, so I guess we'll find out tomorrow, but for now I'm stumped.

I'm going to start pulling memory and CPUs next.

Thanks,
Phil

[auphil]

Well, it's not memory, IDE stuff or 3ware cards haha.

I've got it down to either PSU, CPU, or Mobo. I don't feel like re-pasting thermal compound on the CPUs so I'm just going to wait until the overnight shipment arrives then test PSU, CPU, and Mobo (in that order haha).

....I really hope it's not the Mobo but I'm afraid it might be.

Thanks,
Phil

[kestrel1]

If it is starting up with one PSU, but fails when the others are inserted you could have a short somewhere. May not be the mobo, but it is still possible. Check the connections from the PSU's to the mobo, you might find a short.

[Heebeejeebee]

I had a very similar problem with a HP Proliant server. Luckily it was under warrantee because both PSUs failed. The HP engineer said that it was entirely possible that a PSU had failed sending a spike down through the switching hardware that makes the redundant PSUs kick in and made it fubar. He replaced it and the backplane as well as the PSUs.

HBJB

[umpalumpa]

I have the exact same problem with a server i have, it uses the RM213 model chassis.

Did you manage to get it working again

[SYNACK]

It sounds like the power distribution unit (the intellegent switcher that sits between the MB and PSUs) that is the problem. Try powering it up with each of the individual PSUs and if they all go it is most likely to be the PDU.