I apparently need someone to explain like I'm 5 (years old)

[cw823]

Bought X9DRD off ebay - stolen from front porch
Bought another X9DRD, could not get it to power on. It's now halfway back to the ebay seller.
So I jumped on one of the GA-7PESH2 mobos off ebay.

SAME. EXACT. ISSUE.

Now I did see reference to the supermicro connector on the X9DRD motherboard, and maybe that's all my problem was there (I did not see the article mentioning that fix until I'd shipped it out).

But here I am again, another board, 2x E5-2670, 128Gb (16x8Gb DDR3 ECC Reg). With no memory, or single CPU in either socket, same behaviour. BMC does seem to go through initialization but will not grab an IP address.

I'm stumped, which is admittedly rare when it comes to figuring out how to put together a server. Any and all help appreciated.

 

[Patrick]

Are you not getting any video? Is it getting stuck in the post sequence? Have you cleaned the CPU contact pads? Used a different PSU with both 8-pin power connectors installed. Are there any shorts under the board?

Here is one I learned this week: what type of cooler are you using?

 

[Marsh]

In recent weeks, I purchased 10 complete Intel 2u barebone servers ( lga2011 socket ).
I have my share of strange problems.
Many times, I had to start with 1 single CPU (must be CPU1 - first cpu socket), one stick of memory.
What it helps me is that I used a known good E5-2670 cpu and 1 good stick of memory.

If it doesn't power on, check cpu bent pin, even it looks perfect with a magnify glass. I would still have problem.
Next up, is to spray 90% alcohol into cpu and/or memory sockets, lots of alcohol to saturate the sockets.
Wait for it to dry, or air blow dry. So far , I have got 9 out 10 Intel servers working fine even pass 12 hours stress testing.
1 more to power it later today.

Even if it does not power up, keeping singing the song "I believed"

 

[Patriot]

Are those ES chips? Same chips that wouldn't work in the other boards?
Pads dirty?

Reset the cmos? BMC might be set to a static IP.

 

[ponky]

Check if there are extra stands shorting the MoBo.

 

[cw823]

Are you not getting any video? Is it getting stuck in the post sequence? Have you cleaned the CPU contact pads? Used a different PSU with both 8-pin power connectors installed. Are there any shorts under the board?

Here is one I learned this week: what type of cooler are you using?

No power on. The only activity I get is the BMC LED blinking. I've used 2x PSU, a Supemicro 900W & and a Seasonic x-850.

It's not even in a case at this point.

 

[cw823]

Are those ES chips? Same chips that wouldn't work in the other boards?
Pads dirty?

Reset the cmos? BMC might be set to a static IP.

The are E5-2670 SR0KX

I did yank the CMOS battery and walk away from it for about 15 minutes at one point. I'm not handling it well, lol.

 

[cw823]

Here is one I learned this week: what type of cooler are you using?

I have a fan from a supermicro case fan hooked up to one of the fan ports, I will hook up a fan to both CPU fan headers. Just using a passive heatsink right now but no heat generated as the board doesn't seem to want to power on. Only the BMC? heatsink is warm

 

[Marsh]

No power on. The only activity I get is the BMC LED blinking. I've used 2x PSU, a Supemicro 900W & and a Seasonic x-850.

Yesterday, I worked on a Intel server won't power on. Strip it down to 1 CPU and 1 ram, still won't power on.
CPU socket looked out, spray alcohol , wait an hour to do something else. Plug in 1 CPU and 1 ram. It booted to Windows server.

 

[dataoscar]

Try a different psu?

 

[cw823]

headed to choir (church) now, I thoroughly soaked it in spray alcohol and may say a prayer or two while at church.

 

[Marsh]

Good luck.
Let me repeat myself , 1 CPU in CPU1 socket, 1 RAM in ram socket A1 of CPU1 , always remember to populate ram in blue color sockets before black sockets.

 

[cw823]

I"ve tried both CPU in socket for CPU 0. Only 1 stick of RAM, but i'm not even getting far enough for RAM to be the issue. The board does nothing but blink the BMC LED. Just like the X9DRD that I had last week.

It can't be coincidence. Going to test the CPUs in my other rig, verify they are good. Not sure what else to do at this point.

 

[cw823]

Both CPU test good on another board. All RAM at least posts on another board.

Not sure I can call it anything but DOA. I cannot see any issue with any pins, and I've made my share of money fixing bent pins on 1366 mobos. I suppose I need to source yet another board. Sucks that the issue with that X9DRD that I had was probably due to poor mfgr documentation.

 

[T_Minus]

What is "it" you soaked in alcohol? The CPU or the Pins on mobo or both?

I have these little alcohol squares that I clean every CPU I get with, make sure nothing is stuck on the CPU like a gob of thermal paste, etc...

 

[Terry Kennedy]

No power on. The only activity I get is the BMC LED blinking. I've used 2x PSU, a Supemicro 900W & and a Seasonic x-850.

It's not even in a case at this point.

This may be a silly question, but what are you using to tell the motherboard to power up? Both the Supermicro and the Gigabyte motherboards have a front panel connector which includes a power button (momentary-contact pushbutton). The connectors on the two boards are different, and you'll need the optional Supermicro splitter (normally CBL-0084 or similar) to get individual 2-pin connectors for the front panel functions when using a non-Supermicro board in a Supermicro chassis.

It looks like the power switch function on the Gigabyte board is a momentary jumper between pins 11 and 13 on the 24-pin F_Panel1 connector.

 

[cw823]

This may be a silly question, but what are you using to tell the motherboard to power up? Both the Supermicro and the Gigabyte motherboards have a front panel connector which includes a power button (momentary-contact pushbutton). The connectors on the two boards are different, and you'll need the optional Supermicro splitter (normally CBL-0084 or similar) to get individual 2-pin connectors for the front panel functions when using a non-Supermicro board in a Supermicro chassis.

It looks like the power switch function on the Gigabyte board is a momentary jumper between pins 11 and 13 on the 24-pin F_Panel1 connector.

Yep I've been using a jumper to short those pins, to no avail. nothing ships until Monday so I have some time to find a replacement.

I"ve spray cleaned the pins as much as possible. Both CPU boot right up in my other rig, I actually left the one in place of the 2628L that was in there. Shouldn't be this difficult to make the switch from single 2011 to dual.

 

[Terry Kennedy]

Yep I've been using a jumper to short those pins, to no avail.

Briefly, or are you leaving the jumper in place?

I"ve spray cleaned the pins as much as possible. Both CPU boot right up in my other rig, I actually left the one in place of the 2628L that was in there. Shouldn't be this difficult to make the switch from single 2011 to dual.

I don't put a lot of stock in cleaning the socket pins or the CPU pads. Latching the socket places a fair amount of force on each pin and the connections should be good.

Does either board have IPMI as well as the BMC heartbeat? If so, the IPMI may be able to acquire an IP address via DHCP and you could then connect remotely, at least to see what happens when you tell the IPMI to turn the board on.

 

[abstractalgebra]

Check if there are extra stands shorting the MoBo.

+1, I would remove the motherboard from the chassis and set it on cardboard or similar for testing. Have you tried a jumper direct/shorting on the motherboard power pins or always using the chassis power switch?

 

[cw823]

final testing was on cardboard. I've never seen a device just refuse to power on, but can't think it coincidence that I have the same behaviour across multiple brands of motherboard. memory is good, PSU good, ram is good, nothing else plugged in to the board.