Hi guys,

 

Just wanted some advice about this situation that i'm having.

Data was being rebuilt for one of my disks which i had upgraded to a larger size, it was progressing nicely up until 59% and 4 hours to go then i got some pending sector warning on another disk.

i have only 1 parity disk. Before upgrading the drive i did a parity check to be sure that parity and the other drives were all ok.

There were no problems when doing the parity check, but just my luck that another disk starts to fail when i'm halfway rebuilding data on a new disk...

 

 

so guys, should i just let the data rebuild complete?

 

will the rebuilt disk have corrupted data due to the errors from the other drive?

 

i'm worried that if i stop the data-rebuild now, i'll be faced with 2 failed disks and with no way to rebuild the array.

 

What should i do?

 

 

Please post the diagnostics.

hi JB,

 

here's the diagnostics.

 

earlier i had paused the rebuild for a while... then i restarted and it ran for a while without new errors, but just around a minute ago i got another warning and the pending sector count increased to 4 and now there is 1 reallocated sector

so i have paused the rebuild again.

 

silometalico-diagnostics-20240207-2347.zip

Disk4 appears to be really failing, I would probably still let the rebuild finish so you have more options, you can then try using the old disk to rebuild disk4, but this can only work if parity is still valid, any writes to the array since the disk upgrade?

any changes should still be in the cache because i had stopped mover from transferring anything to the array.

i think the parity should still be valid?

 

i was worried that the data on the rebuilt drive would be corrupted because of the disk errors during the rebuild process.

10 minutes ago, limawaken said:

i think the parity should still be valid?

In that case it should be mostly valid, it can never be 100% valid because the disks were mounted with the new disk, but this is usually OK with xfs.

 

11 minutes ago, limawaken said:

i was worried that the data on the rebuilt drive would be corrupted because of the disk errors during the rebuild process.

There could be some corruption, if the errors coincide with data, finishing the rebuild would be mostly to offer some protection in case the old disk also fails.

i was thinking... would it be possible to switch back to the old disk1 (since all the data inside it is intact)? so i can replace the failing disk4 first.

does that make any sense?

however the new disk1 (whose contents are currently being emulated) is bigger than the old disk1 (4tb vs 3tb).

will unraid accept back my old disk1, maybe just re-validate the parity?

or would the failing disk4 also cause the parity to have corruption when rebuilding the replacement disk4?

22 minutes ago, limawaken said:

does that make any sense?

Yes, that what I posted above:

 

1 hour ago, JorgeB said:

you can then try using the old disk to rebuild disk4, but this can only work if parity is still valid

 

To try that do this, before or after the current rebuild finishes, as you prefer:

 

-Tools -> New Config -> Retain current configuration: All -> Apply
-Check all assignments and assign any missing disk(s) if needed, including the new disk4 and old disk1, new disk4 should be same size or larger than the old one
-IMPORTANT - Check both "parity is already valid" and "maintenance mode" and start the array (note that the GUI will still show that data on parity disk(s) will be overwritten, this is normal as it doesn't account for the checkbox, but it won't be as long as it's checked)
-Stop array
-Unassign disk4
-Start array (in normal mode now), and post new diags.

 

 

 

 

hi @JorgeB

 

sorry, my sleep deprived brain was slow to pickup.

 

the data rebuild on disk1 completed without additional disk errors from disk4, now the array looks to be healthy.

 

 

now instead of swapping back to the old disk and doing all that, i would like to instead try mounting the old disk1 using unassigned devices and copying the data over to the array. i believe rsync would be good for this, right, as it would only copy the parts of the files that are different from the source?

 

so i tried this, using the rsync command

rsync -avh /mnt/disks/X7TY08TAS /mnt/disk1/

 

however all i get is

sending incremental file list

sent 766.30K bytes  received 1.52K bytes  1.54M bytes/sec
total size is 2.90T  speedup is 3,782,919.89

 

there doesn't seem to be any data being compared or transferred.

 

am i doing this wrong?

 

5 hours ago, limawaken said:

i believe rsync would be good for this, right, as it would only copy the parts of the files that are different from the source?

Only if you use the checksum flag, because the file sizes and modifications times will be identical.

 

--checksum
If the file size and time match, it will do a checksum at both ends to see if the files are really identical.

 

Note that it will take much longer because of the need to verify the checksums for all files.

hi @JorgeB, i used the ---checksum flag, and rsync took around 7 hours to go over almost 3TB of data.

i guess 7 hours isn't too bad. by the end of it there were 6 files that were copied. at least i didn't have to swap the disks around and rebuild the data again.

 

i'm starting to suspect that my supermicro drive cage has something to do with these errors that i'm getting. these problems seem to occur after i move drives in or out of the drive cage.

i always do power down before i swap any drives out, just to make sure that the HBA card detects all the disks before unraid starts up.

i had an experience this time when a drive couldn't be detected until i had to sort of install the drive into the caddy in a certain way to make it sit a little further in towards the backplane (if that makes any sense).

 

anyway, thanks again JB, you saved my sanity.