Read Errors during Parity Checks

[Troussdesoin]

Hi there,

 

I've had a few issue for a few weeks, my docker were randomly crashing because one of the SSD in BTRFS Cache was sometimes dropping out.

On Saturday, because of the crash it seemed to have corrupted the Plex docker and its metadata,. So i removed the docker (with krusader, because i could not do it without root privileges) and restarted from scratch.

 Yesterday, i also stopped the server and cleaned the NVME drive to make sure everything was ok and it would stop dropping out.

 

And today i made a parity Check , and i found more than 500 errors on the array. 

I would normally conclude that both Drive are failing, but this seems weird.

 

Is it possible that while removing the files, it would create the read errorrs?

Should i order new drives and replace them, they are only 2 years old. What's the probability of them both failing on the same day?

 

Thanks for the help

unraidchoupi-diagnostics-20200113-0814.zip unraidchoupi-syslog-20200113-0814.zip

[JorgeB]

Both disk 1 and 2 are having read errors, run a SMART extend test on both, there's also a lot of segfaulting and btrfs checksum errors, so possibly bad RAM, run memtest, sync errors could also be related to that, make sure you're respecting max RAM speed depending on config, some Ryzen servers are known to corrupt data with overclocked RAM.

 

[Troussdesoin]

Hey, thanks for the answer, 

 

I ran a Smart extended test last night on one of the drive, and it says it was completed without error. Here is the file attached. i'm going to test the other one and see if the ram is overclocked today.

 

 

thanks a lot.

EDIT: the ram is not overclocked it was a 2133

WDC_WD40EFRX-68N32N0_WD-WCC7K6SKLNHN-20200113-2303.txt

 

edit 2: I uploaded the second smart test

it seems to spew errors especially on the cache drive that was dropping out, i'm planning to change it this week end

 

WDC_WD40EFRX-68N32N0_WD-WCC7K5VY4XY7-20200115-0920.txt

Edited January 14, 2020 by Troussdesoin