Post/BMC problem with old school ML115 G5

Hello Motherboards!!! Mother of all OCF forums....

Right to the point!
Some of you may already be familiar with my rig from:

http://www.overclockers.com/forums/showthread.php/761217-The-White-Elephant

I am not going to review any of the details here regarding that thread except to say:



I have posted in the HP Enterprise forums with no love so far with forty something views and no replies yet.
Don't waste your time reading it as I will re-state the problem briefly here.
You may want to take a look at my avatar over there, hardy har har...
I don't think they appreciate my kind over there?!
http://community.hpe.com/t5/ProLian...-problem-with-old-school-ML115-G5/m-p/6815366

First thing, this board has been out of the case a number of times and has not always done this.
Of course I was careful, but it got knocked around a bit.
I thought I may have knocked off one of the numerous jumpers. Upon comparison to spec docs, I can not see anything wrong.
If there was physical damage, I would think it would cause more evident symptoms, like not working at all.
Instead of this weird intermittent behavior. I have removed/reinstalled since the problem showed-up to make sure there
was nothing buggered up underneath the board.

Two problems:

1.Sometimes Post hangs at the HP Proliant startup logo. Never goes any further. I have to cold reset and then wait!!! If I try again right away, it wont work.
After waiting a number of minutes, it will post again.
If I do anything that requires a restart it will hang again and I have to wait again.
I cant figure out what is buggering it up at the start.
The MLs have a heartbeat led for the LightsOut BMC management, Which is part of their firmware.
I do have a heartbeat before I powerup and while the post is hung. And I do have this error in my BMC event log:


EvM Format Ver=04(IPMI v2) ; Sensor Type =0F (POST ERROR); Dir Type = 71; Data Type 0E00000; Description = CMOS checksum error

I have flash updated the BMC and Bios to the latest available versions for this model.

2.The original OS was w2k3 SBE x86. I want to get a 64 bit os installed. I had previously tried XP x64 and it works wonderfully.
I'd like w2k3 x64, but not sure how to obtain it just quite yet.
But since the POST problem started, when it does get to actually booting up the setup UI after copying drivers/etc,
I get a brief flash of the initial welcome/loading screen, then darkness.
The monitor indicator light stays on, but I get nada.


ALSO, I get the same problem when I tried to run the HP Insight Diagnostics Offline Edition!!??
It gets all going and then when it launches into the UI(linux), flash-nothing.
I was able to run the diagnostics from another utility, that does load its linux UI.


Its almost like the onboard firmware video drivers are hosed??!! Is there such a thing?

Oh, and yes!!!
I have changed the battery.
I have tested/verified the RAM.
I have tried diferent SATA ports.
Every possible way to reset/restart
I ran a complete diagnostics, with no errors!!?? I really wanted an error

And I leave you with that diagnostics report.
Any help here is appreciated gents! and ladies...

Cheers

Have you verified the PSU

Hanging on a restart is typically a motherboard or PSU issue.
Have you built it outside the case w/ bare essentials to test for any possible issues from being in the case?

My thought exactly ATM. Check for any swollen caps on that board. It's no spring chicken

+1 on the PSU
I wouldn't think it would be bad caps already, but if the server was ran as it was supposed to, 24/7 for a few years it might actually be the caps..

A sewing pin works wonders in removing the old solder from the holes where the cap's legs would go on the board. Just got to get it hot

Johan45 said:

Have you verified the PSU

Click to expand...

err, uhh yeah. Its there and on...

ATMINSIDE said:

Hanging on a restart is typically a motherboard or PSU issue.
Have you built it outside the case

Click to expand...

not since I posted/tested it with the new psu w/o anything connected.
Posted right up, but that was before I ripped into this thing.
Could be the next step...

Johan45 said:

Check for any swollen caps on that board.

Click to expand...

...

Silver_Pharoah said:

24/7 for a few years it might actually be the caps....
..works wonders in removing the old solder

Click to expand...

So, yeah, replaced the factory 386W psu with:
FirePower OCZ Fatal1ty OCZ550FTY 550W
http://www.newegg.com/Product/Product.aspx?Item=9SIA4H81E87157

I figured it gave me necessary juice for the h80i gtx AOI and some room for OC.
So I assume the PSU is not the problem here?

Caps??? Swollen?!?
You mean like capacitor caps? or whatever those VRM components are by the cpu?
And they would actually be swollen?
The rig got use for maybe 12 months active service and the LightsOut allowed it to totally sleep/shutdown when it wanted to.
And you mean i could actually replace, if the problem?

Here is a thought to munch on!!!
The issue that I have to wait and cold reset to post again...
I thought this had to be some capacitor retaining charge or something.
So I would do the usual click and hold power button while unplugged. This does not seem to do the trick.
WHat does do the trick is this:

-While hung at post logo click the power btn one short click. System will shutdown after a few seconds.
-Heartbeat and MB power led stay lit.
-Click off the power on the PSU. (leave it plugged in)
-Walk away, have some coffee, smoke a ***, drink a beer, do drugs, whatever. Just dont try to restart immediately.
- comeback, click on psu, wait for mb led and heartbeat indicator, click power and bingo it posts.

I got sick of trying, waiting 30 seconds, then trying waiting 60 seconds. SO I just walk away and do something else.
SO I cant be sure of exactly how long it actually takes, but on the order of minutes.
Could this be some buggered component that is not discharging? based on these weird symptoms????

As always, thanks for the insight and help...

Cheers

Yes, swollen. The lids on the capacitors should be flat. If they don't look flat they could be on their way out. Also if the sides are bulging bit, that's a sign.

See if you can get some nice clear pics of the board again.


FWIW, I had similar symptoms on my PIII server here. It was the PSU. The main capacitors were bulging so I bought a new PSU from Oaklahomo_Wolf. All is kosher on that server now, apart from the BIOS glitches lol.

err, updated shots?
Here are the originals...
low res, you can kinda zoom in.
I can get some more once I yank it to bench test it.
What types should I be looking at? The red/silver ones, the mercedes symbol ones, all of them?

Thanks

What's this thingy

So what is this fun looking group of components?

I ask because, if you look, it's where the mb/bios LED is and those mercedes capacitors next to that copper wrapped deally are what I would classify as jiggly.

Torrids are what the copper wrapped jigglers are

If there are any bad caps, it'll be the Mercedes ones
Not often you see the solid capacitors bulge.

Also check for shorts anywhere. Metal touching screws, shorted pins etc.

SwartHack said:

Here is a thought to munch on!!!
The issue that I have to wait and cold reset to post again...
I thought this had to be some capacitor retaining charge or something.
So I would do the usual click and hold power button while unplugged. This does not seem to do the trick.
WHat does do the trick is this:

-While hung at post logo click the power btn one short click. System will shutdown after a few seconds.
-Heartbeat and MB power led stay lit.
-Click off the power on the PSU. (leave it plugged in)
-Walk away, have some coffee, smoke a ***, drink a beer, do drugs, whatever. Just dont try to restart immediately.
- comeback, click on psu, wait for mb led and heartbeat indicator, click power and bingo it posts.

I got sick of trying, waiting 30 seconds, then trying waiting 60 seconds. SO I just walk away and do something else.
SO I cant be sure of exactly how long it actually takes, but on the order of minutes.
Could this be some buggered component that is not discharging? based on these weird symptoms????

As always, thanks for the insight and help...

Cheers

Click to expand...

That is the mark of a ****ty 5vsb on the PSU.

Mr. Scott said:

That is the mark of a ****ty 5vsb on the PSU.

Click to expand...

What is an "sb"?

Ok, I got the guts yanked and everything on The Bianch.


I am also using a different PSU.

SilverStone Strider Plus ST50F-P 500W
http://www.amazon.com/gp/product/B0047CQ1B4

This was recently yanked out of BruceLee for a cabling upgrade and I know it is good.
Well actually, I really don't "know" anything at all...

So same post problem on The Bianch.
I can live with this for now, especially if I can get a dang 64bit os installed.

So any love on problem 2.
If you look at the picture you will see where I am hung on an XP64 install.

"Setup is starting Windows"

I found this pretty extensive(old) thread about it but none of the solutions work for me.
http://www.hardwareanalysis.com/content/topic/13533/

Interesting thing is that before I jumped into this project and ya'all were helping me figure out exactly what I had

I had installed XP64 without problem when I had the original w2k3 SBE x86 os running????
but I was booting from the CD so it shouldn't have mattered what other os was there...
COme to think of it I think I unplugged the w2k3 drive when I did it.

Why can't XP setup see anything. What did I do?
besides flash/update the bios...

What is APIC exactly anyway? Disabling it does not work for me.
When I use the F5 trick I get two choices: ACPI uniprocessor x64, ACPI multiprocessor x64
so I assume I need ACPI enabled?

I can see back in the day this was a problem, so maybe some of you old-timers may have some insight.

Salud

Probably missing drivers for something on that motherboard. If it has some not so common controllers then Windows may not have a driver for it.

5vsb = 5 volt stand by

APIC = https://en.wikipedia.org/wiki/Advanced_Programmable_Interrupt_Controller

You want APIC enabled, and it has to be done before you start to load your OS.

Uniprocessor if you have only 1 CPU
Multiprocessor if you have more than 1 CPU
Cores don't count either. It's number of sockets.

XP64 blows for driver support. That's 90% of your problem. You also need to supply the SATA drivers at F6 on install to see your drives.
Why are you not using W7 on this?

Johan45 said:

Probably missing drivers for something on that motherboard. If it has some not so common controllers then Windows may not have a driver for it.

Click to expand...

Yeah, more on that in a second. I still am perplexed as to how I installed XP64 in the first place.
I realized I had the original HP factory seagate spinners stashed away.
One has the original factory w2k3 SBE x86 install. The other has the XP64 install I originally performed.
Neither one would boot up. Just as either get to firing things up I get a BSOD.
Says something like windows has stopped to avoid damaging your hardware or something like that....
I think it may be due to the bmc and bios flash updates I have performed since using those disks last???
I cant go backward on the firmware that I know of???

So, this could be really fubar, but I am not one to give up so easily...

Mr.Scott said:

...
XP64 blows for driver support. That's 90% of your problem.
You also need to supply the SATA drivers at F6 on install to see your drives.

Click to expand...

bwahardy har, when you hit F6 it says:

"Insert Floppy disk in A: drive. Enter to continue."

There is no option to browse, nada..

Why are you not using W7 on this?

Click to expand...

ahh yes my intrepid celestial itinerant, let me refresh your cyberspace logs...

In my initial inquiries into what the hell I actually had, which you Sir where very instrumental in:

RJARRRPCGP said:

BTW, Windows Server 2003 is so *not* a total POS. XP x64 is based on it and that would be the OS to use unless you upgrade the RAM. (to 5 GB or more)

If it's got 4 GB or less, I would change some DDR 2 sticks.

And anything Vista and later may have non-existent driver support.

Click to expand...

This ml115, is obviously end-of-line at HP. The latest supported firmware/drivers are for w2k8 R2.
And, if you also recall the whole "FInding my PLL" fun we had and that everything on this rig
is integrated into the ServerEngine chip, which was made by Matrox. And uses nVidia drivers,
INCLUDING THE IDE_SATA DRIVERS!!!




So I am basically limited in my options here.
FYI - I did get w2k8 R2 to install and win7x64 as well, they work OK. But the CoolCPu, setFSB nor the moc (Matrox OVerclocker) worked, which was not surprising.

So where are we now?

I have been trying to build an iso with nLink that includes everything I need, specifically the ml115 nVidia IDE_SATA package.
I managed to do this easily and the install actually goes further now!!!
Instead of just hung up, I get a Blue Screen!!! Yay!
Well at least it tells me "IRQ less than or greater error, something like that, in usbport.sys!!

So I go out find a hot fix for w2k3_XP_64. Include it in my iso build and same error
That is where I am. I am going to plug on with nLink and see if I can't get something to install.

In the meantime, any insight is always appreciated!

Prost

POST this up your BMC

I guess my lesson for the day is that persistence is the key.

I just couldn't accept what was going on and was trying every possibly conceivable approach:
-hours under the mag lens, top and bottom, scrutinizing every connection. cautiously checking every component for wigglies.
-20 different cd burns of different OS's
-and something like an attempt at 10! Bios setting combinations.

And on top of this I had to wait ~10 minutes just to change a Bios setting due to the POST problem.

So in the middle of it all I mentioned nLite and got into that. What cool software. I will never load drivers after install again.
I realized that the OCF ISO library where I got the original XP64 was made with it, duurrr.
I interloped on their forum and ran into RJ over there. He was no help though.

Hey RJ!!

One thing I kept running into researching 13 year old forum topics was the issue of mouse refresh is XP/w2k3 x64 and how to fix it.
I was at first, could gives a rat licking buttt about mouse refresh rates. But they kept coming up and kept addressing USB speed.

Ding Ding Ding

Bios Setting->Advanced->USB Configuration->USB 2.0 Controller Mode=FullSpeed (12Mbps)

Default was HiSpeed (480 Mbps)
and apparently XP/w2k3 x64 do not have any way to use the HiSpeed, hence the mouse rate issues..
-
So reset CMOS and changed this setting, and only this setting and

BINGO

and guess what, it also fixed my POST problem automagically!!!??? Not sure why/what how??

One thing I think, but cant really verify is that the 7/2009 BMC and BIOS versions (Last official release for ml115)
I used to flash with maybe added this setting and/or changed the default value.

When the install started and I could format/config disks I fell out of my chair.

The kids will be so thrilled!

Hasta Entonces

Funny how BIOS's are eh?

I can hear the kids cheering now!!

A little follow up...

Mr.Scott said:

XP64 blows for driver support. That's 90% of your problem. You also need to supply the SATA drivers at F6 on install to see your drives.
Why are you not using W7 on this?

Click to expand...

Well Sir, I finally did fingure it all out and have win7x64 running with all the appropriate system drivers.
It came down to all the reps and research from fixing this problem that I became all to familiar with all the pieces.
I then stared to research Win7x64 for driver versions of the individual pieces and basically blew off the HP support site.
It came down to the nVidia System management package HP used and what exact pieces I needed/required.

So I am about off to the races with this rig.

Also, kinda funny story, saved for the kids, but
I'll recycle a little here from:
http://www.overclockers.com/forums/showthread.php/766851-Novice-Nimble-6



Pic of testing new board with sempy, jacked into WhiteElephant.
The stock HP cpu fan is just sitting there, without being screwed down.
The Proliant case does not use a backplate, but posts mounted on the case underneath the mobo.
I need to hackeneer some posts on the bench so I can try to mount the predator to the sempy.




Thanks everybody for all your help and putting up with me

Brazos Fuertes and Feliz Ano Neuvo

Glad to hear it's up and running smoothly now
Nice lil treat with the sempy, OC the beans outta it!!