Guys sorry. I'm not a new user, and I'm not tech illiterate; but I don't keep up on the ins and outs of uR . In fact...that's one of the things I like best about it; it's a great product, that can mostly be set-it and forget-it. Until there's a problem...and then, the community here is always great at helping out.

 

Just started noticing some write errors; and I snapped this, in the middle of a parity-check. What do you think it indicates exactly...and what should be my game-plan?

 

TIA

 

See here

See here

 

Thanks...attached.

New_Text_Document_3.zip

Post  SMART report for disk5. See here: http://lime-technology.com/wiki/index.php/Troubleshooting#Obtaining_a_SMART_report

The example uses sda. Your image shows disk5 is sdb but this can change so check the current letter value before getting the report.

Post  SMART report for disk5. See here: http://lime-technology.com/wiki/index.php/Troubleshooting#Obtaining_a_SMART_report

The example uses sda. Your image shows disk5 is sdb but this can change so check the current letter value before getting the report.

 

Attached. Thanks

smart.txt

SMART overall-health self-assessment test result: FAILED!
Drive failure expected in less than 24 hours. SAVE ALL DATA.
See vendor-specific Attribute list for failed Attributes.

 

Replace the drive ASAP.

Is there a new method...to reduce array, and just remove the drive for now?

 

http://lime-technology.com/wiki/index.php/FAQ_remove_drive

Is there a new method...to reduce array, and just remove the drive for now?

 

http://lime-technology.com/wiki/index.php/FAQ_remove_drive

 

OK; having real trouble, understanding how to copy files off the bad drive? 

Have you got a spare disk ready to copy the data to?

Have you got a spare disk ready to copy the data to?

 

I probably could, yeah; but I thought you just copied it back into the array? (if you have the room, which I do)

Yes you do, copy the files from the faulty disk/s to one or more of the others. And then replace the faulty disk. It looks like you may have two faulty though.

Yes you do, copy the files from the faulty disk/s to one or more of the others. And then replace the faulty disk. It looks like you may have two faulty though.

 

OK; wow. I'm copying one user dir...from the bad disk 5; to another networked drive. Saying "1 day" remaining...lol (~900G).

 

This is necessarily? I can't just re-align the array...to exclude that disk in the re-build; and then pull it?

If you want to leave the drive out of the array after copying the data off the drive/s you can stop the array, shutdown, pull the drive/s and do a new config on restart. Make sure you assign the parity to the correct drive. Take a screen shot before so you know the serial number of it. Once they are all reassigned start the array and a new parity will build. Be aware if you do it this way you have no protection for another drive failure.

Copying a full TB (900G is close) is simply going to take time.    Since the disk is faulty, you do NOT want to do a New Config and exclude it -- you may need to let UnRAID reconstruct some of the data from the other drives and parity to successfully finish the copy.

 

A couple questions ...

 

(a)  Did the parity check you noted in your first post complete okay?    Does it show zero sync errors?

 

(b)  Are you copying the failing disk's data to another disk on a different PC, or to another folder on your UnRAID server (on a different disk) ?

 

Note that the errors shown during the parity check are not write errors -- they're all read errors, that were corrected by UnRAID by reconstructing the data from the other disks + parity and then re-written (successfully) to the failing disk.    You DO want to replace this disk, but not until you're sure you have a good backup of the data.    Alternatively, you could simply replace the disk and let UnRAID rebuild it onto a new disk ... but only if your parity is good (i.e. the check finished and had no sync errors).

 

 

^^ OK, let me try and answer some of these questions; as I want to do the safest...but also fastest fix. In other words...I just don't want to do more than I need to; at the expense, of days of back-up.

 

1st; parity check was good

 

 

Next...I am currently copying, the data from disk 5 to another drive; outside the array.

 

3rd...I realize disk 4 is also starting to fail; but I think I still have some time, before that's completely useless...and I'll deal with it soon enough.

 

Last...for now; I am planning to replace a drive. So I know I could just do a "replace", and save myself this copying(?); but it might be a few days...and since I had the room in the array...I just thought I could move it along?

 

Somewhat OT...what would you vote for these days: WD Red or Seagate NAS? I'll order a 3T, right-quick.

Yes, you can simply replace disk 5 with a new 3TB drive and UnRAID will rebuild the data onto the new drive.    No copying necessary.  HOWEVER, I'm a firm believer in backups (as I suspect you know), so I'd let the current copy complete before you do that.  THEN I'd replace disk 5 and let UnRAID do the rebuild.  As long as that's successful (it should be), you won't need to copy the data back.

 

So ... the process I'd suggest is:

 

(1)  Order your new 3TB drive, but let the current copy finish (it'll be done before you get the drive)

(2)  If you don't mind taking the extra day or so to do it, I'd pre-clear the new drive as a confidence check on the new drive.    This doesn't speed up the rebuild, but it will confirm you have a good new drive.    This is, of course, optional -- the simply fact of doing the rebuild, followed by a parity check, will be a very good test of the new drive.

(3)  Replace drive 5 with the new drive, and let UnRAID do the rebuild.

(4)  Do a parity check to confirm all went well.

 

Done :-)

 

r.e. which drive to buy ... both the WD Reds and the Seagate NAS are excellent choices.    The Seagates are marginally faster (5900rpm vs 5400rpm), but both work very well.    I use nothing but WD Reds in my array.

 

I agree with garycase's method, however I would not leave disk4 to long before replacing and be aware If you copy the data off the failing disk5 as you are doing, then replace it with a new one and rebuild it and disk4 fails you will lose the data from that. If you have the space on the server or externally I would copy that data off disk4 as well as a precaution. As its going to take at least 33 hrs to preclear a 3TB drive, if you want to preclear it, I would. See what garycase thinks. I never worry to much as I have offsite backups of all my data. And like garycase I use WD Reds.

Agree that copying the data from disk4 as well is a good idea ... "just in case".    It's always a good idea to have backups of your data (both of my arrays are fully backed up).

 

r.e. your comment:

...  I want to do the safest...but also fastest fix

 

... I always focus on SAFE.  Speed is secondary when it comes to ensuring the integrity of your data -- ESPECIALLY if you don't have backups to fall back on.

 

The fastest thing to do in your case would be to simply replace disk5 and let UnRAID do the rebuild.  But with another "iffy" drive (disk4) that's taking a risk that the rebuild finishes before disk4 fails ... so I wouldn't do that without a backup.

 

I'd simply do as I outlined above ... possibly being even a bit safer by also backing up disk4.

 

Thanks guys. I feel like I'm in good hands...as always; and I'm ordering one drive tonight.

 

While copying from disk 5...I've already gotten the message, that one of my files (a movie) was inaccessible. Am I going to lose it...or likely, the re-build will work (off parity)...and the copying step is more of a precaution?

It's more of a precaution because disk5 hasn't red balled your not using the parity to emulate it. As long as your parity was good and disk 4 doesn't fail during the rebuild you should be okay.

Actually I'd say you're going to lose the file that isn't accessible.  Since parity was perfect (no sync errors), the disk, when emulated, will be exactly as it is right now.    So if the file system is corrupted for that particular file, it will be the same when emulated.

 

You MAY be able to recover it with reiserfsck, but that's another process entirely.  Wait until after you've replaced the disk -- but do NOT do anything to the old disk -- and then you can try reiserfsck if there are any files you want to try and recover from it.

 

My mistake I though he said his last parity check was good showing no drive errors. I suppose it depends on if he's written to parity or not?

The fact the parity is good is why I don't think a reconstruction will help.    If the file system was somehow corrupted, the same write that did that would have also resulted in parity being updated, so the reconstructed drive will have exactly the same file system corruption as the actual disk.    What MAY help is a reiserfsck, which can correct many errors of this nature.    But I'd do this one step at a time -- get everything backed up; then the drive replaced; and then try reiserfsck.

 

OK...drive is here, and I should be done copying the last of the data (at least what I could still access) by tomorrow.

 

So...I stop the array, replace the bad drive (disk 5 for now); start-up, bring the array back online...and do a "re-build"?

 

Working...

OK...drive is here, and I should be done copying the last of the data (at least what I could still access) by tomorrow.

 

So...I stop the array, replace the bad drive (disk 5 for now); start-up, bring the array back online...and do a "re-build"?

 

Working...

 

OK, so...I backed-up all the data from disk 5 that I could (I did get some errors, that certain files could not be accessed; obviously the sectors that were failing), replaced the drive; did a rebuild and did a parity check. Parity check found and corrected 187 errors. What's that mean exactly? Is that an indicator, of what was lost moving from the bad disk to the new one...or errors with the also-going-bad, disk 4 (which is currently at 3293 writes and 579 errors)?

 

I know Gary had said he thought I would lose some data...because my parity returned with no errors; so...is this somehow "better"? You know, in hindsight...it was a total waste of time, to copy the old data before the replace and rebuild. I have a SATA to USB dock! That would have worked, right?

 

OK...so where do I go from here? For now, the array is up and running; I know I need to do something about disk 4 now (I'll post a smart report), and maybe that's why I "lost" files in the rebuild? BTW...with sooo many files; how are you supposed to know, what's "gone" and what's not...until you happen to try and use something, and can't get it?

 

TIA