Parity errors, but drives are healthy

[Jarlehm]

Hi all!
For the last couple of parity checks I've been getting errors. I'm suspecting two older 8tb drives in the system, but I can't figure out which drive (if any?) is giving the errors. My box has been running fine for about 18 months. I was going to expand the storage anyway with a new 16tb drive, currently waiting for at the post office, but it might be better to replace one of the old drives with the new one? 

torchwood-diagnostics-20221002-1242.zip

[JorgeB]

Bad RAM is the main suspect of unexpected sync errors on a server without ECC RAM, start by running memtest.

[Jarlehm]

Thanks JorgeB, I'll look into that. Haven't heard of ECC RAM before, but from what I can gather, it's probably better to change the RAM I already have to ECC regardless?

[itimpi]

3 minutes ago, Jarlehm said:

Thanks JorgeB, I'll look into that. Haven't heard of ECC RAM before, but from what I can gather, it's probably better to change the RAM I already have to ECC regardless?

You can only use ECC RAM if the motherboard supports it.

[Jarlehm]

Ah... I was minutes from pulling the trigger on new RAM, there. My motherboard doesn't support it. Guess that's an upgrade I should invest in down the line. Thanks itimpi! 

I'll run the memtest now and see how it goes.

Edited October 2, 2022 by Jarlehm

[Jarlehm]

Memtest has been running for almost 17 hours now (9 passes). No errors thus far. Is that long enough? 

 

Since the RAM doesn't appear to be the issue, what would be the next suspect to check out? 

[JorgeB]

Next suspect would be a disk, are you sure the previous check was correct?

[Jarlehm]

I wouldn't claim to be sure of any thing😅

If there is some way for me to mess it up, it is definitely possible, but as far as I can recall, I haven't changed anything about how it runs except the schedule. 

 

The previous parity check also had some errors (around 600), so I guess something isn't right. 

 

The memtest is still running without any errors, by the way. 

[JorgeB]

25 minutes ago, Jarlehm said:

I wouldn't claim to be sure of any thing

If you are not sure I'd wait for the next parity check, last one was definitely correct, so if the next one finds errors there's still a problem, if it doesn't you are likely fine and in that case the 1st one was likely noncorrect.