I got a notification that I had read errors on disk 4 (226 errors on the main page, and a raw value of 39 on the "Raw read error rate" ). I did a short smart test that passed, then a long test which failed after 20%, dunno why, maybe the read errors (?). I then did a parity check and got 0 errors, but the read errors had then increased to 396 on the main page and 82 on the Raw read error rate. I then powered of the server and booted up again (read someplace that a power cycle might help (?). Did a new smart long test which failed again after 20%.  Now after I booted up again the errors was ofc gone from the main page. But my question is should I replace it now or should I just watch it and if I keep getting errors replace it then? I'm guessing I'm gonna see errors again because the smart long test failed?

tower-diagnostics-20160120-2024.zip

I don't see any serious issues with your drives, but do see that a couple of them have non-zero raw value for:

 

UDMA_CRC_Error_Count

 

This is typically an indicator of a cabling problem (bad or loose cable) to that drive.

 

Note that the value will never decrease, so after you fix the problem it will not increase, but will not return to zero. So I have no way to know if the problem was fixed in the past, or is the cause of your read errors. But seems a possible cause to the symptom you are reporting.

Ok, I'll check the cabling and hope it's a loose or bad cable.  Great excuse to open up my server and plug in my spare disk and preclear it which I was supposed to do a long time ago!