Did I dodge a bullet with this non-ECC RAM?

[rezin8]

Over the weekend I upgraded the hardware in my FreeNAS box from a Core i7-4790K with 32GB of DDR3 to a dual Xeon E5-2660 with 128GB of ECC RAM. I put the old hardware on a test bench and ran memtest86 on it as a quick test. This is what came back:

I know about @cyberjock's thread on ECC RAM and many of the other discussions on it, which is one of the reasons why I upgraded, but did I really dodge a bullet here? Was my pool really at risk?

 

[Mirfster]

Well, having Non-ECC Ram in itself can put the data/pool at risk. So I would have to say that having that same non-recommended RAM and it having error(s) would not be good either..

Side note, you said you upgraded over the weekend; hopefully this was after the new equipment was already burned-in before being implemented...

Never a good idea to just upgrade components in a production system without first testing them out fully.

From the jgreco's "Building, Burn-In, and Testing your FreeNAS system" thread:

This process also needs to be repeated when you are upgrading your system by adding new parts or changing out existing parts. There's a little bit of "use your brain" in how strict you need to be, but doing stuff like just dumping more RAM into your box and then booting it with a production pool attached can lead to great sadness. Don't do that. Your hardware needs to be solid and validated, and if it isn't, you can scramble your bits, possibly irretrievably.

 

[rezin8]

Well, having Non-ECC Ram in itself can put the data/pool at risk. So I would have to say that having that same non-recommended RAM and it having error(s) would not be good either..

Side note, you said you upgraded over the weekend; hopefully this was after the new equipment was already burned-in before being implemented...

Never a good idea to just upgrade components in a production system without first testing them out fully.

From the jgreco's "Building, Burn-In, and Testing your FreeNAS system" thread:

Yes, the hardware sat on the test bench and I hammered it for quite a while before transplanting it into the chassis.

 

[Mirfster]

Good man :D

 

[Ericloewe]

Over the weekend I upgraded the hardware in my FreeNAS box from a Core i7-4790K with 32GB of DDR3 to a dual Xeon E5-2660 with 128GB of ECC RAM. I put the old hardware on a test bench and ran memtest86 on it as a quick test. This is what came back:

I know about @cyberjock's thread on ECC RAM and many of the other discussions on it, which is one of the reasons why I upgraded, but did I really dodge a bullet here? Was my pool really at risk?

Not really. That's a really tough test that does not represent any realistic workload. It's more of a potential hardware vulnerability to malicious software.

https://en.wikipedia.org/wiki/Row_hammer

Even the developers (which are somewhat dubious) thought they needed to tone it down. Almost all machines will not have zero errors on that test. It's really not indicative of bad RAM.