Unraid - Preclear failures? - HELP

[NeoMatrixJR]

I'm still running 6.4.1 for now...I had a boot drive failure the other day and recovered from that other than a few corrupt docker templates.  Now I've had my first actual drive failure in the array.  I was getting a couple of disks prepped to add to the array when my 3tb drive hosting much of my data failed. I tried running preclear on the two drives I'm adding (used, from other systems) and they both fail post-read.  I've read that preclear isn't necessary anymore, and some skipped the pre/post reads.  I've managed to move all data from the failed 3TB drive to other drives in the array (disk is in emulation). SMART reads fine for both "new" disks (a 2TB and a 4TB) except yellow banners on UDMA CRC error count...can I add them to the array?  Do I need to worry that they failed preclear?

 

I'm running extended SMART tests on them now.

[Frank1940]

You are asking about four different issues in one paragraph.

 

First thing, I would suggest you do is to post up your diagnostics file.   Tools   >>>  Diagnostics   (Hopefully, you haven't rebooted since you started having problems with hard disks in general.)

 

About preclear, it isn't necessary if the disk(s) you are adding are good.      IF they are bad, the preclearing can save you a lot of headaches.   Only you can decide where you like to be dealing with a 'bad' disk... mounted in your array or in the preclear operation.   

 

Normally, UDMA CRC Errors are not an indication that a drive is 'bad'.  Since, they are a count of the number of times that the data that went into one end of a STAT data cable and came out the other end different.  This is type of errors is usually either a bad cable or a bad connection.  Occasionally, a bad SATA card can also cause these problems but these are usually accompanied by other indications that the card has problems.  

[Kode]

What was the failure in the post read?

[Sean M.]

Were you using the pre-clear plugin?  If so which version and which script and operation.

[NeoMatrixJR]

Attaching diagnostics....

 

Am I missing it or is there a further log on the preclear?  I'm wondering if there's any more information than just that it fails....  This goes to @Kode - Sorry...I started writing this before you responded.  I'm not sure what the failure was...it just says failed.

 

@Sean M. - Preclear script: gfjardim - 0.9.5-beta.  Plugin version is whatever one CA installs by default right now.

theconstruct-diagnostics-20180427-1108.zip

[Frank1940]

It seems like there are some serious issues with disk2 (sdh) that is identified as ST3000DM001-9YN166_W1F0HPDV.  In fact, I can't even find a SMART report for that disk in the Diagnostics file.   

[JorgeB]

4 minutes ago, Frank1940 said:

It seems like there are some serious issues with disk2 (sdh) that is identified as ST3000DM001-9YN166_W1F0HPDV.  In fact, I can't even find a SMART report for that disk in the Diagnostics file.   

 

I'm guessing that was the failed disk.

 

Both unassigned disks look fine, they both have a very high number of CRC errors, unless these are old you need to replace the SATA cable, other than than there should be no issues using the 4TB one to rebuild the failed disk, and it doesn't need to be precleared for a rebuild, the 2TB won't do as it's smaller than the original.

[NeoMatrixJR]

sdh is indeed the failed drive.

 

I assume the CRC errors would carry over (stored on disk) from previous use.  They've only been plugged into this machine for a couple days and they're now on my r710's backplane.  One was in my desktop PC for YEARS, but I was noticing some issues with that drive recently...May have been loose/bad cable recently...   The other was previously in an external USB case (HGST TOURO drive).  The drive quit working and I shucked it for diagnosis and possible re-use.  Seems like the USB controller went bad IMHO.  The drive wouldn't even come up before, but no issues once plugged directly into sata.

[NeoMatrixJR]

Is there a log or part of the log that details why a preclear failed?

[NeoMatrixJR]

found 'em.  Only interested in logs pertaining to sdi at the moment.  Logs still don't tell me much...either I don't know how to read them (good probability) or they don't detail the problem well enough for me to spot it.

THECONSTRUCT-preclear.disk-20180503-1255.zip

[John_M]

Read failure:

May 01 07:43:02 preclear_disk_PL1331LAGJT30H_19813: Post-Read: dd command failed, exit code [141].

 

[NeoMatrixJR]

Yeah...I can't find much past that.  Dd failed 141...what does that mean?  anyone know?  I haven't found a good answer to that yet.

[Squid]

1 hour ago, NeoMatrixJR said:

Dd failed 141...what does that mean?

Quote

When a command terminates on a fatal signal whose number is N, Bash uses the value 128+N as the exit status.

 

Which seems to imply that the process exited on SIGPIPE 

[trurl]

You really need to be concentrating on rebuilding the failed disk.

[NeoMatrixJR]

On 5/6/2018 at 3:55 PM, trurl said:

You really need to be concentrating on rebuilding the failed disk.

I pulled all data off that disk while it was ghosted and moved it to others.  Figured I couldn't rebuild until I had a reliable disk.

<facepalm> 

Turns out preclear has since been updated due to....dd failing post-checks....

Updated preclear and started a 2-cycle on my 4tb disk.  So far it's passed the 1st cycle fine.....

[trurl]

1 hour ago, NeoMatrixJR said:

I pulled all data off that disk while it was ghosted and moved it to others.

I usually discourage people from moving data from the emulated disk to other disks in the array since it is just a lot of additional activity on all the array disks with an unprotected array. Rebuilding from the emulated disk would get your data back, and you will have to rebuild it even if you moved all the data off. I think it is better to just copy the emulated data to an unassigned device or other machine on the network if you can't immediately rebuild.

[NeoMatrixJR]

Fully agree...I wouldn't have pulled from the emulated disk if I knew for sure I was going to have a good disk to work with.  I was worried I may have another failure before I could fix what I had.  I may consider purchasing a drive, and pre-clearing it for an emergency spare before I fill all my r710's slots.  Thanks again for the help!