[They indicate a problem... possibly with the disk electronics, but more likely with the disk controller or system RAM.

 

The script  wrote zeros.  In those locations indicated, it read back 32768.    If you use that disk in your array you will likely pull out your hair with constant parity errors as the values read back from the disk are not those written.

 

Since the returned value always seems to be a power of 2, I suspect a marginal "bit" in the electronics.  (or system RAM)

 

I'd start with a memory test, through several cycles, to ensure it is not RAM, followed by a systematic analysis of the remaining hardware.  Based on the report that you had the same issue with the prior drive, it is likely either the disk controller, or the system RAM.

 

I have replaced the RAM in the unraid server and it has passed 15 cycles of the memory test with out any errors.  Memory was upgraded from 512mb to 2GB.

Joe, I have previous posts about weather or not the -A is required in the pre-clear script.  I was concerned that with out the -A option, it tells me that the partition will not be aligned.  Can you take a look at my previous posts on this thread and give me your opinion?

 

Thanks for your time.

 

Sincerely,

 

Sideband Samurai

Don't be concerned.  It will set a proper signature on the disk when a disk is > 2.2TB, regardless of the option you supply (or do not supply)

 

instead of continued posting, just use -A. 

(and it will be ignored AND SILENCE the warning about sector 63 which DOES NOT APPLY when a disk is >2.2TB.)

Joe and everyone;

 

Thanks for all your feedback, Everything has been resolved.  Ultimately it was RAM in the server causing all my concerns.

 

-- Sideband Samurai

Ultimately it was RAM in the server causing all my concerns.

 

Memory can cause a large variety of issues -- and is also one of the more frequent things to fail.  I suspect that's why Tom includes it by default on the flash drive's setup -- and makes it a simple boot-time choice 

I guess I have one more question that fits here.

 

When pre-clearing disks, how many cycles should I run on the drive before it is considered good.

 

For each 4TB drive I pre-clear each cycle takes 24 hours.  If I run 5 cycles that's 5 days before I can put the drive into service.  I want to use best practices at this point to make sure the hardware is sound.  Just wanted feedback on what people do.  When I originally put the system together, I ran only one pre-clear and the drives have performed flawlessly. 

 

Thanks.

 

- Sideband Samurai

 

Most errors are caught in the first cycle is what the majority of the people will probably tell you.  I use 3 cycles myself.  I don't trust just running 1 cycle.  If I could afford the time I would probably use 6 cycles.  It would have made finding a bad 3TB WD Green drive easier when it went from 65535 to zero to 65535 to zero, etc... pending sector count.  It was suggested to me that it had a bad firmware.  That particular drive had at least 6 cycles before I was done diagnosing the smart errors over a 3-6 month period.  It started out new with 128 after 1st cycle, 32 2nd cycle then zero 3rd cycle when I put it into service.  Got some sync errors after a few months and smart said there was pending sectors again.  So I took it out and ran some more preclears to clear them.  Ran at least 3 and that was when I noticed the alternating pattern on pending sectors.

you should put that drive on an ebay auction.

you should put that drive on an ebay auction.

I would never stick someone with it.  But sending it back to WD might do the same thing as they might just recertify it.  I probably have waited too long to send back to WD although I believe it was still a 3 year warranty when I bought it so maybe not.

I meant to list it explaining the bizarre behavior. Bizarre people may bid it up. Check the WD website for warranty coverage. You just need to enter the SN.