PowerEdge 2900 E1410 CPU IERR was asserted

Hi David,

 

Thanks for your reply.  I brought this machine at Feb 2008, so I believe it is under maintenace.  I'm not sure if it's hardware failure.  The error message showed when the room tempture is higher than 30C.  However it works fine at 28C.  Is it a normal condition?

 

Jaguar 

No, it is not a normal condition. It is a legitimate error.

I also believe there are something wrong in PE2900.  However I call the servcie guy in Dell, they told me that it's only becasue of the higher temperature and everything is ok.  Is there any other tools that I could check the potential error of PE2900?  Please advise.

 

Thanks for your help.

 

Jaguar 

This error occurs when the CPU asserts its Internal ERRor (IERR) pin.

 

There are several possible reasons for this. There's a couple of listed processor errata that can cause this, but your failure to boot in what are not unreasonable ambient temperatures is of concern.

 

You could adopt a 'watch and wait' approach. However, if it happens again, I'd want an engineer out certainly to reseat the heatsink for that CPU, and really to change the CPU for another. As snapohead says, it's a real and legitimate error.

PE2900 died today after I got a message E122C this morning (System Board CPU Power Fault: Voltage sensor for System Board, state asserted was asserted).  It could not be started anymore.  Dell engineer will replace a new CPU and MB.  Becasue of the weekend, I need to wait until next Mon.  :( My question is why it failure?  Do you have any experience for that?

 

Regards,

 

Jaguar

 

Hardware failure can easily be just one of those things. It could be that something like the voltage regulation components for the CPU failed on the motherboard; the first sign of this might be some sort of blip that caused the IERR.

 

A new motherboard and CPU should get you going - how unfortunate it failed on a Friday so that your service call isn't until Monday.

Dear David,

 

Many thanks for your recent comments.  I just a little depression, because I thought the PE2900 should be very stable.  Is there any suggetion about the environment for PE2900?  I do not want to have a hardware failure again.  Or any suggetions for the redundant system? I do nothing without this server. :(  

 

Thanks again for your help.

 

Jaguar

 

It's probably just "one of those things" - you could join the service engineer on Monday to see if there's an obvious component damage on the motherboard near the CPUs. I wouldn't be surprised if there's nothing to see, though.

 

There is a fair amount of redundancy in these servers (typically they have RAID disks and redundant power supplies), but it isn't possible to duplicate everything.

You could buy another server and set them up in a cluster for increased redundancy.

---

@snapohead wrote:
You could buy another server and set them up in a cluster for increased redundancy.

---

Indeed! Those single points of failure will always catch you out.

 

If you are going to cluster, the greater the separation between the two servers, the better. For example, if you can connect the second server to different network switches, you've added redundancy against a network switch going down. The same applies for power and UPSes. It's also best, if possible, to put the two servers in different locations in case there's a problem that affects a single room or single rack.

 

Of course, the amount of separation may be limited by your software - as with so many things, it's a balancing act.

 

 

Hopefully you're back up and running after the service engineer's call today.

It works fine after replacing MB, CPU, MEM.  One of the MEM also failure and could not be detected after replacing a new MB.  Because they also brought the PERC/i and DRAC, I asked them to replace all to make sure everthing is ok.  The service of Dell is good.  Only because of the weekend, I could get the support until Mon.  And also I appreciate your comments for my problem. :)

 

 

Jaguar 

Great - it's good to hear that all went well with the engineer call. It sounds like something fairly significant went wrong to need a new CPU, motherboard and at least one stick of memory. I agree that it was wise to replace as much as possible.

 

By the way - though it's really an academic question - does your 2900 have one CPU or two?

 

 

Hopefully that's the end of the problems. If not, I'd be pushing to get the power supply (or power supplies) replaced. If there are any further problems, I'd also check out your power to the machine.

 

However, I suspect it was one of those things, and all will be well now.

I have only one CPU, 4G RAM (2x2G), 500G RAID 0 (2 SATA HDD).  I'm interest in why it is a academic question?

 

BTW, I will continuous to monitor, if it still problem, I will use your suggestion to replace my power supply.   Of course I really hope it is the end of problem.  Thanks again for you help.

 

 

Jaguar 

One tip for stability's sake - if you really meant RAID 0 and it wasn't a typo, I'd aim to discontinue the use of RAID 0 as soon as possible. RAID 0 doesn't give you any redundancy and isn't, in my opinion, suitable for server use outside specialist applications that have a reason for needing it (such as some video editing setups). RAID 0 is less robust than a single hard disk - if either disk fails, all the array data is lost.

 

A pair of 500GB hard disks are relatively inexpensive - that will allow you to have a 500GB RAID 1. If either drive fails, the array will keep on working with a single drive until you replace the failed drive to restore redundancy. That said, RAID 1 is not a substitute for proper backups.