Curious RAM Controller Failure ?

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

Well, bit of a ball ache this one.

Dell Poweredge 600SC, 512mb DDR ECC Registered

Customer Description: Voicemail software/server freezers and requires rebooting.

Diagnosis indicated failed ram, failed Memtest v3.5, v4.0 and the Windows Mem tester, although the internal diagnostics utility gave it an all clear, the system ended up hard freezing in all three above tests, with memtest giving an "unexpected interrupt - halting cpu". It would be the only type of RAM that we didn't currently have a spare stick of lying around, so an order was placed to test.

Today: New 1Gb stick inserted (old was 512mb), tested, failed, head in hands, wtf O.o. Did some more tracking the noted that all the errors on the 512mb stick occur at 511.9mb, all errors on the 1Gb stick at 1023.9mb (nb few low end errors ~1mb). Now thats just too much of a coincidence.

1Gb: 1023.9mb / 3fffdc80 (error location)
512mb: 511.9mb / 0001fffdc80 (error location)

The rest of the system passes it's internal utility checks, cpu caches ok, timers ok, everything "appears" fine.

I'm scratching me head here and having a well deserved coffee after spending hours hacking the image of this system into a VM from a clonezilla image yesterday. Ultimately I'm going to suggest they virtualise the box anyway as it's noisy and needs replacing with something newer but this is bugging me from a technical perspective.

Anyone come across this before, any ideas? There's a fresh pot of dark roast on if you want some

 

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

did you run memtest on one dimm at a time in all slots for several passes?

If not then do that. Memtest is only truly accurate on one stick.

Doing that will either prove the ram is faulty or the slots are faulty.

If you have done the above then I think it would be safe to assume the slots are faulty.

Mixing ram makes and speeds can cause problems and even some motherboards don't like different sizes.

 

Ensure that the memory slot isn’t clogged with dust or anything that would cause connection issues.

Try the stick in a different memory module bank then try running mem test to make sure its not a physical fault (If the board will boot like that)

Make sure the RAM speeds are configured correctly

Try updating the BIOS

 

Unable to test in other slots, server refuses to post unless there is a stick of RAM present in slot 1.

Multiple tests run against it, all fail, cleared the slots of dust (whilst doing the usual loud coughing) previously.

Quoting this morning on a virtual box, if we end up with the hardware I might do some more digging out of curiosity.

 

If you cant't run without a dimm in slot one then memtest maynot be entirely accurate but if its bring up loads of errors then I think its the slots at fault and not the memory. Try cleaning the connectors and contacts with a rubber from the end of a pencil and see if it helps.

 

Ive had issues before similar to this, but not had the time to find out 'exactly' what the problem is. We had some issues where a mobo would only show 8-10GB even though 12GB was installed, mem test would test 12GB, Bios would recognise 10GB etc etc - and ended up just replacing components until the problems fixed.. This sounds to me like its the motherboard playing up. It's unlikely that a brand new stick of memory is fauly unless your very unlucky.

Is there another machine you can test the newer memory module on?

I know its a long shot but have there been any changes to the servers normal environment? Power cut / Surge? Forceful shutdowns? Have it been moved / Transported?