topic Redundant power supplies, but the server failed anyway. in ProLiant Servers (ML,DL,SL)

Redundant power supplies, but the server failed anyway.

g0lem — Fri, 25 May 2012 20:02:01 GMT

Hi, All. I guess this is a question about setting expectations. My customer bought a Proliant ML-350 g5 server. She has the redundant hot swap power supplies plugged each into it's own battery backup. Earlier this week, the server crashed. She had an amber power light and the LED on one of the power supplies was dark. She was able to re-start the server. I was able to unplug, and re-plug in the failed power supply and it showed green, the amber light went out, & I ordered the power supply anyway. Last night it failed again, and again, the server crashed.

While it's certainly handy to have a spare power supply installed so it's there when the other one fails, I thought the idea was that the server would not crash! because of our clever redundant power scheme. Is this expectation incorrect? Is there something else we should be doing? This customer requires and is willing to pay for actual high-availability. Please advise, and thanks for sharing.

Re: Redundant power supplies, but the server failed anyway.

chinmay — Fri, 25 May 2012 21:05:24 GMT

Well there are lot of issue seen with ML350 G5 PSU. Please confirm the PSU part number is 403781-001

Ideally If 2 working PSU is installed in the server the led should be green for both. Yeah your expectation is correct and it should not crash. It very much can be the board between the psu and the system board (power backplane I guess).

However, the power backplane part is usually passive so it doesn't fail often by itself (maybe a broken PSU caused it to go bad).

Try the below steps to conclude if the incoming power / UPS/cable are not the cause

As you said She had an amber power light and the LED on one of the power supplies was dark, which PSU had no light bay1 or bay2? Have you checked that the power actually reaches the PSU? Try swapping the psu and and confirm if the issue is with the bay or goes with particular psu

Try connecting both the PSU directly from the wall outlet, and unplug one while its running, check if the server shuts down.If you have confirmed rest everything works, you might have to replace the psu backplane. If the PSU part number doesn't match with the below one or doesn't have any yellow dot you can replace them as well.

Just for reff Part Number for ML350 G5 Server
403781-001:Hot-plug power supply, 1000W, 12V
413144-001:Power supply backplane
Link: http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=125&prodSeriesId=1121586&prodTypeId=15351&objectID=c00727361

If my Post is useful Please Acknowledge by clicking the white KUDOS Star!!!!!!!!
-Chinmay Kabi Satpathy

Re: Redundant power supplies, but the server failed anyway.

g0lem — Fri, 25 May 2012 21:58:28 GMT

Hi Chinmay,

Thanks! Your post is encouraging. That's not quite the part number. The part # on her server is 413982-001. I was able to unplug one of the power supplies will no interruption earlier. When I bring out the new one, I'll try your troubleshooting systematically.

Re: Redundant power supplies, but the server failed anyway.

chinmay — Fri, 25 May 2012 22:23:29 GMT

Thats sounds great, let me know if I could be of any more help. BTW part number 413982-001 is for Front bezel - For use with tower servers.
Check the spare part list link below for ML350 g5 and do CLT F for the same number, which will give you the description.
http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=125&prodSeriesId=1121586&prodTypeId=15351&objectID=c00727361
as it a hot plug power supply...unplug the psu and you can see the number on it.

If my Post is useful Please Acknowledge by clicking the white KUDOS Star!!!!!!!!
-Chinmay Kabi Satpathy

Re: Redundant power supplies, but the server failed anyway.

Goutham_Sabala — Mon, 28 May 2012 14:10:17 GMT

Needless to say G5 servers especially ML350 G5 servers have quite some issues with power supplies and psu backplanes , the first thing I would do is to replace the psu with 403781-001 ( latest psu ) earlier ones ( 379123-001 or 379124-001 )-affected ones. Then monitor the server and then decide if psu backplane needs replacement.

Re: Redundant power supplies, but the server failed anyway.

Giorgio911 — Sun, 23 Feb 2014 13:41:22 GMT

I have a ML370 G5 that was turned off for a scheduled power outage, and now it will not turn back on, and the green LEDs on the hot swap power supplies will not come on?? Paralyzing a 22000 member credit union...

Re: Redundant power supplies, but the server failed anyway.

waaronb — Mon, 03 Mar 2014 23:19:55 GMT

If neither supply is working, it kind of sounds like your power line itself is faulty. Trying different outlets or bypassing the UPS to see if it's on the fritz could be helpful.

Hopefully by now, a week later, you've figured it out and a follow-up to see what the resolution was would be interesting, to satisfy our curiosity. :)