Just added a new drive to my server (not in the array yet, just been preclearing it). Was just checking the disk log and noticed a number of errors popping up: 

 

i.e. Jul  7 12:01:51 Tower kernel: blk_update_request: critical medium error, dev sdm, sector 5555444736

 

The preclear was successful, I haven't added the drive to the array yet. Should that error be something that I should worry about. The drive is still in warranty so could look to return it if needed. One of my drives has a number of reallocated sectors, but the error above is appearing on a different drive.  

tower-diagnostics-20170709-1220.zip

SMART looks fine, replace cables or swap backplane just to rule them out and run an extended SMART test.

I would not ignore an error like this. With a new drive, I am always suspicious of an install issue, and would double check the cable connections.

 

I really urge users to use hot-swap style drive cages so that adding and exchanging drives is not accompanied by this type of drama.

2 minutes ago, bjp999 said:

I would not ignore an error like this. With a new drive, I am always suspicious of an install issue, and would double check the cable connections.

 

I really urge users to use hot-swap style drive cages so that adding and exchanging drives is not accompanied by this type of drama.

 

Luckily I do have a hotswap back planes on my D8000 so can easily swap it to another bay without too many issues. 

 

Will give it a go later today. 

 

Would it be worth running another preclear on it in a new bay? Or just an extended SMART test? 

Either one is a good test, if using preclear make sure at least one of the read options is selected.

The extended drive test is good for testing the drive, not the cabling, controller, backplane, etc.. The preclear tests the drive in its installed context. So I tend to prefer a preclear. If you continue to get errors, an extended test would be a more definitive diagnosis of a bad drive.

I'm running a preclear on it at the moment. Will report back.