Two parity check runs, both finished with 1 error.

[aspik]

Hi,

I want to replace a smaller disk with a new bigger one. To be sure that everything is OK I started a parity sync (write corrections to parity enabled). The parity sync finished with one error. I then started second run of parity check (also write corrections to parity enabled). After an approximately an hour one error was found/corrected. This second run of the parity check finished with just that one error.

How can two parity checks result with one (same?) error?

How can I find the root cause? 

In the past I never had parity errors. Recently I moved the server to a new case, could it be a cable problem?

Please find attached diagnostic logs.

 

Thanks in advance for any help.

fractal-diagnostics-20240423-2203.zip

Edited April 23 by aspik

[JorgeB]

One possible explanation would be the first time that bit was wrongly correctly, like from a bit flip, so the second pass would correct it again, but you appear to be using ECC RAM, so that should not happen, possible an issue with one of the devices, run another check when possible to see if more errors are found.

[aspik]

Yes, I'm using ECC memory. Ok, then I"ll start a third run. Will report in 24h about the result. Thanks for looking into it!

[aspik]

The third run finished with 0 errors. Attached logs after the check completed. Can I trust the parity? It is safe to do a disk swap, or should I check something else before exchanging the disk?

fractal-diagnostics-20240425-0954.zip

[JorgeB]

1 hour ago, aspik said:

Can I trust the parity?

I would, at least for now, just keep monitoring for the next scheduled checks, but if more errors appear it it should be investigated.