HP ML350e Gen8 Fan Noise

[Digo79]

3 years later and the issue is still unresolved.   This thread has 31261 views!! 

I've had a case open for a month now, and HP hasn't been able to come up with a solution... Come on HP, is anyone still in charge of the Gen8???

[CardinWest]

I ran SPP 2015.10 on two DL380p G8 servers and the fans started going nuts. Fan 5 and 6 (CPU1 Side) were running between 70 and 80% off and on. After checking the ROM Power Profile for Balanced and the Power Regulator for Dynamic I thought it could have been upgrading the firmware from P70 08/02/2014 to P70 07/01/2015 but the fans kept going after I swapped the ROM. 

To cut to the chase - it was the StarTech USB 3.0 card. I tried the 2 port and 4 port. Either way, once the card is removed the fans settle down.

One of the DL380pG8 servers is the dual CPU 25 SFF bay and the other is a single CPU 8 bay with a P420 for an additional 8 bay cage. I don't know if the extra heat on that side of the box make fan 5 and 6 rev up from the cards but I have a feeling it was the iLO upgrade from 2.20 to 2.30. That was the only thing I did not downgrade but I don't need the 3.0 ports that bad. I love HP and I love StarTech but in this case the two do not mix. Had no problem with the G7s in the stack.

This took me about 8 hours over two days to figure out so I hope I save someone some time and heartache

[Torsten.]

In such case it is always a good idea to connect to ILO, go to component status page and check the temperatures.

Some time ago one of the peers was testing something in hardware configuration - some minutes later the server powered off because of overheat. The reason was the opened cover, so the fans could not cool down the disks, the disks reported 62 degrees celsius - the server powered off 'randomly', depending on the temperature of the disks and made a lot of noise...

[PGTRI]

hi,

For those who are using 331 NIC. Try to downgrade the FW to following version:

Windows:

http://h20566.www2.hpe.com/hpsc/swd/public/detail?swItemId=MTX_cbcf4954bd744f44977d2fa95d#tab3

ESXi:

http://h20566.www2.hpe.com/hpsc/swd/public/detail?swItemId=MTX_ef8b4a00061f46c188821fa9f5#tab4

Please share the result.

thanks,

[renefl]

My fan sppeds were constant until I upgraded to SSP 2016.04.

Downgrading the 331 NIC firmware to the one in the link solved the problem.

Thanks a lot!!

[James MacMaster]

We experienced the constant ramping up and down of the fans on a ML350p Gen8 after updating to SPP 2016.4.0

Fan speeds had been fairly low and consistent prior to the update.

Downgrading to the 331 NIC firmware in the links provided by PGTRI resloved the issue.

Appreciated thankyou

James

[reddogchop]

Excellent answer, after upgrading to the latest SPP (ML350p G8) fans 2 and 3 were ramping to 18% and as high as 33%; downgraded the NIC firmware per link and hey presto all back to the usual 6%.  Nice work..

[hpesxi]

I get 

[root@localhost:/tmp] ./CP023981.scexe
This version of ESX is not supported. Exiting,,,

Which version of ESXi were you using?

[hpesxi]

I unplugged the thermal cable and the fan goes from 31% to 22% idle.  Still too loud but better, something is wrong here. 

[cincinnatimann]

ack this probem is due to few things, and not just to one,  fisrt remove all heatsinks off mother board clean all chips and reapply a really good thremal compond,  age of pc and kind of themal compond used drys out, and u get chips running much hotter, now to cpu if one look at heat sink one find few probems what made of not copper, size of heat sink to small for amout of btu the cpus are giveing off, throw even said this, is fix, note looking at chip on endge of chip there gold places,. this must be insulated with well anything dose not conduct elec, the breack that holds down cpu can short these places out, now to themal past there is a themal past is a liquid mettel, be very carefull it is and will conduct elect and 10 times heat, chip must be clean and i do meen clean and so dose heat sink in fack best thing clean heatsink with is accatone, u only put enouf to cover the chip no more make shure all corners are cover do not put more then is requred to cover the chip, in shot less the better, reinstall heat sink,  i have rebuilt many of these, and be honest heatsink is the probems, that and rest chips over heating,  this should make fans slow down 50% if done right if not u blow the chip,  this stuff been on market very long time and works all so well but this is liqued mettel is nothing to take lightly it must not run off the sides of chip, again do not add more then is needed, if u do all this work it will run better then new < u can do same with video card will reduce heat by factor of 30%

[AndyChips]

We've had a similar problem with a DL380 Gen9. Once every 5 or 6 months it goes into meltdown. All fans on 100% and the server (Hyper-v 2016) is unresponsive. The only thing to do is to power it off.

The only other time it happens is when a network broadcast storm happens, i.e. when someone loops an ethernet cable back into the same switch.

Even though it happens infrequently I'd love to find a fix for the first issue though.