Anyone any good with Dell servers ?

[Big Jim]

We use a 2nd hand PowerEdge 2900 in our shop.

Today noticed it was noisy.
Fans are spinning at full speed and the front panel is warning about a CPU Temp sensor, code E1118

Done a bit of googling around and not come up with much.

Have tried clearing the error log from from dell server administrator console.
have tried swapping the CPUs over (2 x Quad core Xeons)
have tried running with just one CPU (tried both)
have updated BMC from 2.37 to 2.5
have check bios nothing in there.
if I shutdown and pull the power then boot back up it returns to previous state for a moment
ie, boots up at full speed, slows fans right down

this doesn't last though in the space of 30 seconds or so it boots the fans back up to full speed in 3 stages.

Not really up to speed with servers if I am honest so not sure where to proceed from here ???
It is out of warranty obviously.

 

[angry_geek]

Grab an infrared thermometer and check the temps. If they're ok, then you could be on the way to a dead motherboard, or the sensor could just be gone. Sometimes, reflashing the bios will kick it back into submission. If you determine it's just a faulty sensor, you could grab an aftermarket fan controller to bring the speed and noise down. However, if it's the sensor or you go the fan controller route, the system could start shutting itself down thinking it's overheating. A replacement board, depending on which one you have, should run about $150 or so. . . if you can find one.

 

[knc]

Dell has/had phone support for life on older servers, you can try calling in with the service code and I bet you get support. You would of course have to pay for the parts but they can diagnose it for free.
1-866-687-3355 Dell support

 

[nlinecomputers]

Dell should have diagnostics that should report the temperatures that the sensors are reporting. I'm betting a faulty sensor. Other thing to look for is reported fan speeds. If one seems low yet is running at full speed then you may simply have a faulty fan. Some Dell diagnostics can test fan speeds and you can watch and see if they slow down and speed up as commanded.

 

[Big Jim]

Fans are all reporting correctly.
Speedfan reports all temps around 20 - 25oC on all cores (will only display 5 cores however due to all 4 RAM DIMM temps being reported)

Fans are all proprietary hot plug connectors so a manual fan controller would not really be a viable option.

This is obviously done at hardware level as the fans are spinning up to full speed during POST

 

[Bigtek1]

I found this if hope it helps

http://en.community.dell.com/support-forums/servers/f/906/t/19274188.aspx Possible resolution is firmware update.

 

[Big Jim]

tried that (as in my OP) and it didn't work.

Tech has spoken to Dell, they didn't even have that code in their list. they just know E11** errors are thermal related.

Unless anyone else has any other bright suggestions it looks like I am going to have to stump up for a replacement board, have found a couple on ebay for about £80.

 

[angry_geek]

Jim, I don't know how much shipping to UK is (I thought you were USA), but here's some contact info for a lot of parts. He just quoted me $85 for a board for a PE 2950.


Christopher*O'Neill

System Design Advantage, LLC

5653 Creekside Parkway

Lockbourne, OH 43137

Phone:**952-703-3566

Fax:* 614-409-9393



Edit: damn auto correct!

 

[OfficeTechs]

i remember something about having to clear the NVRAM on Dell servers to clear errors occasionally.

Think its either Ctrl+F or Ctrl+E when in the BIOS, thats what it was on a 1950 anyway.

Edit: Also a little tip apparently is to remove the suspected faulty component, switch on the server and wait a few secs then power off, then replace the component and switch back on. I had to do this with a SAS RAID card that was causing an error.

 

[Big Jim]

i remember something about having to clear the NVRAM on Dell servers to clear errors occasionally.

Think its either Ctrl+F or Ctrl+E when in the BIOS, thats what it was on a 1950 anyway.

Edit: Also a little tip apparently is to remove the suspected faulty component, switch on the server and wait a few secs then power off, then replace the component and switch back on. I had to do this with a SAS RAID card that was causing an error.

The issue with doing that is that the fault (appears to be anyway) is a temp sensor, which is built into the board and is managed at a hardware level that even the dell windows based software cannot see (at least it isn't reporting CPU temps in windows)

so it looks like a board replacement unfortunately


Thanks for the offer angry_geek, looks like by the time I add shipping it is going to work out the same price to buy it over here. Thanks for your help though.