1

My HP ProLiant DL180 G6 died on me during the night. Sadly during an raid-5 to raid-6 rebuild so my data may or may not be lost. That's next but first, the server won't boot.

Problem details:

  • Front LED is solid green, but it always is until POST anyway.
  • BMC indicator LED on the motherboard is flashing green.
  • Fans spins to maximum.
  • Power supply LED is green on the back side.
  • No VGA output.
  • No Num Lock response on keyboard.

I have tried the following:

  • Pulled all RAM but one stick - nothing
  • Tried different RAM - nothing
  • Re-seated the CPU - nothing
  • Pulled the HBA controller - nothing
  • Pulled the TPS board - nothing
  • Pulled power to all disks - nothing
  • Different power supply - nothing
  • Tried every setting of the J26 jumper. Clear CMOS and Recovery BIOS had no effect. Recovery BMC made the BMC LED go solid green and the fans span even faster but then nothing and RTC reset turned on UUID LEDs, didn't spin the fans and then nothing.

Any ideas what could be the problem? Faulty CPU? Faulty motherboard?

  • 1
    Dead systemboard I suspect, shame it was during moving from R5 to 6 - but at least your intentions were good - we hate R5 here on SF. – Chopper3 Jul 20 '16 at 13:01
  • @Chopper3 Oh, moot point. Redundant power supplies, redundant harddrives but the motherboard gets me. Perhaps get two servers and mirror them. – winny Jul 20 '16 at 13:12
  • Yeah - depends on your uptime commitments, sometimes just having a single spare for each component to hand will get you all you need - if you need 'fine-nines' or whatever then yeah, clustering is your friend, but then often you have to introduce another bit of kit that is then a single-point-of-failure, like a SAN, to support the cluster - then you end up with two SANs - it all adds up and needs supporting, sometimes just saying 'hardware failure, be back up in 20 mins' is enough and saves a lot of faff – Chopper3 Jul 20 '16 at 13:35
  • 1
    @Chopper3, I'm thinking low tech. Two servers in two different locations which are mirrored. If one fails, repoint the DNS to the other one. – winny Jul 20 '16 at 13:45
  • @Chopper3 New mobo, new thread: serverfault.com/questions/792549/hp-se1220-refuses-to-boot – winny Jul 28 '16 at 18:46
1

Realized I never answered my own question here. After much troubleshooting, I measued the individual voltage rails and realized +5 V was missing. Turns out the PSU backplane does not just MOSFET-diode-OR the voltages from the power supplies together since the PSUs only produce +12 V to start with and the +5 V buck converter failed the high side FET to open.

The raid rebuild was a disaster. So much data lost...

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.