topic Re: ML370 Blank Screen but Fans on in ProLiant Servers (ML,DL,SL)

ML370 Blank Screen but Fans on

StuExplore — Thu, 18 Feb 2010 12:41:05 GMT

Hi, I have an HP ML370 G3 server. This morning it was moved and since then it will not boot.

All the fans fire up, the HD lights come on and all LED's are green, but nothing on the screen, keyboard doesn't get any lights and the USB mouse doesn't show any power either.

The drive are in two logical drives
2 are mirrored and 3 are raid 5.

The mirrored pair showed two arrows underneath for quite some time, these have now gone off and the raid 5 drives have arrows under them.

The fans appear to be on full.

The server was very dusty but after it failed to boot it has been cleaned out, but this has made no difference.

Any ideas?

Thanks in advance

Re: ML370 Blank Screen but Fans on

Erdogan Temur — Thu, 18 Feb 2010 13:41:29 GMT

Check the memory/memory configration and try running with a single memory.

Re: ML370 Blank Screen but Fans on

Manuel Boosch — Thu, 18 Feb 2010 15:35:12 GMT

has to be faulty memory or faulty cpu!

Re: ML370 Blank Screen but Fans on

StuExplore — Thu, 18 Feb 2010 16:06:22 GMT

Tried different memory config and I think the memory is ok. I get beeps when I take it out etc. so it appears to be fine as I don't get any beeps and get green lights when it's all in.

I don't have a way of testing the CPU, but I have taken it out, blown the dust off and reseated it, but this made no difference. My gut feel is that there is a failure on the system board, but it's odd that all the LED's are green.

Thanks for the ideas so far.

Re: ML370 Blank Screen but Fans on

Erdogan Temur — Fri, 19 Feb 2010 07:19:16 GMT

Do you have a motherboard to external devices attached?

Re: ML370 Blank Screen but Fans on

Manuel Boosch — Fri, 19 Feb 2010 08:57:36 GMT

before one week i had the same issue but with an proliant dl380 g6... all lights green and blank screen, after changing and testing the second processor socket on the system board was defective. try a look on the pins for the cpu socket if they are damaged or twisted.

do u have an old ILO card installed ?

Re: ML370 Blank Screen but Fans on

StuExplore — Fri, 19 Feb 2010 09:13:30 GMT

Thanks for all the replies. We don't have anything attached except internally we have a raid controller - which obviously we need.

I've not tried the processor in the second socket - can you do that even without a processor in the first socket?

I did take the processor out yesterday and gave it a dust, all the pins looked fine.

My gut feel is that the motherboard has failed.

Re: ML370 Blank Screen but Fans on

gregersenj — Fri, 19 Feb 2010 10:04:58 GMT

This is a classic :)

The fault syptom is classic.
And a server that has been running for years, is powered down and won't restart.

There can be several HW failures, that caurse these symptoms.

1. Memory.
2. system I/O bd.
3. CPU.
4. Power supply
5. I/O bds and peripherals.

Memory:
One of the very first tests in the POST, is the first 8 kb of memory.
The x86 architecture use the fist KB of the memory for interrupt vectors, so thats im portent. And it's testet before the video/graphics is initialized.
If it fails, the post will stop with no video.
And it's the most fequent reason for this failure.
Could be a bad memory module in socket 1 (1 based count), Improper seatet or bad connection due to dirt/dust.
It could be another memory module caursing short circuit of the memery bus.

Solution Clean the machine (Will get back to that later) Test modules in pairs in socket 1 and 2 (Refer to mem config on the inside of the lit.)

2. System I/O bd.
Lots of components can fail. You will need to try to replace it.

3. CPU. Could fail. Power module for the CPU colud fail.
Yes you must have an operational CPU in socket 1 !!!
Moving it to socket 2 and leave socket 1 empty is a very bad idea on the Proliant servers.
This is due to that the Proliant will disable a bad CPU and reboot (ASR) in case of CPU failure.
When you remove a CPU the Proliant will treat it as failed. So you have now disabled the first CPU.
Normal replacement procedure for a CPU, other than the first, is to replace cpu, then mark it as repaired in the IML and reboot, this will re-enable the CPU.
You can't boot without the first CPU, so you need to clear the IML.
Generally it's not recommened to clear the IML, since you will loose the failure history.

But now you have to do it, using the switch on the system I/O bd. See Label inside the lit.

4. Power Supply.
There's a power good signal on the PSU, wich hold the reset for the CPU. For as long as the power good signal isn't set teh CPU is reset.
So if you got a failure in the power good circuit. you might have a good 5 and 12 volt on the connectors, but no power good signal. (I have no idea which pin it is).
Power supply / power sub system It happens rarely.

5. I/O bds ect. causing short circuit, or hold the reset signal low.
or a bad video card, but in this case it's on the system I/O bd.
Remove all peripherals.
Beware the Prolaint got interlock switch on the I/O cache. So no I/O cage or bad connection, will caurse no boot/POST.
It will give you a red LED on the system I/O bd.

That was the classic failure symptom.

The other classic.
A server has been running for several years. It's powered down or have power loss from the out side.
There's a risk that the power supply(s), or Hard drive(s) won't restart.

If the server is very dirty, you should clean it.
Take care of ESD !!
Disassemble the server.
Use a vacume cleaner with a brush.
Then compressed air for the rest.

Reseating parts, can be the solution. espcially after it has been moved.

BR
/jag

Re: ML370 Blank Screen but Fans on

StuExplore — Fri, 19 Feb 2010 10:18:47 GMT

Thanks for the awesome and detailed reply. I'll give the options a go.

Thanks

Re: ML370 Blank Screen but Fans on

gregersenj — Fri, 19 Feb 2010 10:30:15 GMT

You're welcome

good luck

/jag