Help determining what's up with 2nd parity drive

[SudoPacman]

Hi guys,

 

After a little support here please, since this is the newest drive in my machine, and "has read errors" so has been disabled.

I have replaced one set of SAS cables the other month, but there is another set that could be replaced if warranted.

 

I'm confused in that you don't seem to get much information about what unraid thinks the issue is, other than has errors so has been disabled, unless I am just missing it from the logs or diagnostics.

 

It looks okay from a SMART point of view, which worries me from an RMA point of view.

It also went through the complete pre-clear process without issue.

 

Any help appreciated.

 

Cheers,

Pacman

 

monty-diagnostics-20230501-1020.zip

[JorgeB]

Syslog is incomplete and missing the disk getting disabled, but SMART looks fine, so first replace/swap cables to rule that out and re-sync.

[SudoPacman]

Thanks. Just ordered some new cables, so should have them in tomorrow.

 

Cheers

[SudoPacman]

Okay, so changed cables, and the drive has been through a complete pre-clear, with pre-read and post-read and all looking good.

This is what happened last time though...

 

I'm going to add it as a second parity drive and let it build.

I suspect it'll all go fine, and then next month during the scheduled check it'll fail, but let's see!

 

You said the syslog did not contain any information on why the drive got marked as having read errors. Is this because it did not go back far enough?

Anyway to increase how far back is recorded perhaps?

 

Cheers

[SudoPacman]

Ah, just found the mirror syslog setting. Will try that!

[SudoPacman]

So, no additional failures yet, so that's good.

 

Last time it happened when the parity check kicked off at the start of the month, so will report back with how that goes, and with the syslog capturing enabled.

 

Cheers

[SudoPacman]

Got an update on this in case anyone else experiences the same thing.

 

In a nutshell, it kept happening. I'd get through a preclear, then add parity and boom, at some point during the rebuild.

This is on a practically new supposedly enterprise class 16TB drive, so I was suspicious.

 

A mate at work suggested it could be PCIe lanes, since I'm using an older Intel i7 970 and X58 platform, and during a rebuild there is a lot of strain on the system, and I have a lot of drives (10 data plus 2 parity and 2 SSDs).

 

I've been thinking about an upgrade for a while, and have now pulled the trigger on an AM5 based build on an X670.

Expensive, but it's crucial to me that I can trust my server.

 

Just finished a parity rebuild, and all seems good.

 

Hope this helps someone else.

 

Cheers,

Pacman