drive replacement question - 5 tb to 6 tb to 5tb

[FrozenGamer]

Can i replace a 6 tb drive with a 5tb drive?  That way i can put my 6 back on as a hot spare in case of a failure.  Seems 6tb drives are pretty expensive at the moment.

[trurl]

Can i replace a 6 tb drive with a 5tb drive?  That way i can put my 6 back on as a hot spare in case of a failure.  Seems 6tb drives are pretty expensive at the moment.

Parity doesn't know anything about filesystems, all the disks are just bits as far as it knows. Rebuilding will rebuild every bit of the disk regardless of what is on it. So unRAID will only allow you to rebuild a disk to a disk as large or larger than the original and no larger than parity.

 

You could preclear the 5TB, add it to a new slot, copy the data from the 6 to the 5, then New Config without the 6 and rebuild parity.

[FrozenGamer]

What is the easiest way to copy data from 6 to 5? i am assuming i would not want it to be part of the array or it would possibly get other data onto it?

[tdallen]

I usually telnet into my server, use Screen, and then Midnight Commander (mc) for large disk to disk copies.  Use the disk shares, i.e. /mnt/disk1 to /mnt/disk2.

[FrozenGamer]

Ok, i have found the MC app on my telnet to server (the gui version).  I haven't used it before, but is this the the correct idea/plan?

1- i have a mover from my 5tb cache set for 3:40 am, disable mover schedule if it appears this will not copy in time.

2- add precleared empty 5tb drive to array - assuming it will stay empty if my cache is storing any new data.

3- copy entire contents of 6 tb drive to empty 5tb drive with MC, stop array and pull out 6tb. (would i not move versus copy?)

4-figure out how to use new config (i have found the following reference to unraid 5 stuff which also refers me to this as a more up to date version http://lime-technology.com/wiki/index.php/Shrink_array ) -

"On the 5.X series of unRAID[edit]

Stop the array by pressing "Stop" on the management interface. Un-assign the drive you wish to remove. On the Utils tab you'll see a New Config option. Invoking it will set a new disk configuration. A side effect will be the immediate invalidation of any prior parity calculation.

When the Apply button is clicked, old parity data will be immediately discarded. When you next start the array, the process of parity calculation on the remaining assigned and working drives will begin. At this point, your array will not be again protected from a disk failure until the system can complete the process of generating new parity information."

 

Is it possible to forgo all this and just pull the 6tb (with 4.5tb space used) out, replace it with the precleared 5tb, then rebuild the 5tb as the 6?)

 

Or essentially with the suggested plan would i be moving (versus copying) all the data off of the 6tb i would like to remove and make back into a hot spare, then doing a new config to restart the array without the 6tb in the array?

 

My other option is to introduce 8tb seagate drives harvest from usb enclosures into the array, but that scares me a bit as the only stuff i have read on the forum is using internal 8tb seagates so until i hear more i am waiting.

 

[FrozenGamer]

I also found this which seems to apply and says it is still accurate - https://lime-technology.com/forum/index.php?topic=37210.msg343941#msg343941

[itimpi]

 

Is it possible to forgo all this and just pull the 6tb (with 4.5tb space used) out, replace it with the precleared 5tb, then rebuild the 5tb as the 6?)

Unfortunately no.  You can never rebuild onto a smaller drive (although a larger is allowed).

 

Or essentially wMy other option is to introduce 8tb seagate drives harvest from usb enclosures into the array, but that scares me a bit as the only stuff i have read on the forum is using internal 8tb seagates so until i hear more i am waiting.

i agree I have not heard anything about enclosure drives being used in unRAID.  I use the 8TB Seagate archive drives very successfully, but do not know how they compare to the enclosure drives.

[trurl]

There are several ways to do this but in the end if you want to end up with a 5TB instead of the 6TB you are going to have to rebuild parity. The main question is whether you move/copy the data from the 6 to the 5 with both disks in the array, or with the 5 in the array and the 6 mounted out, or with the 6 in the array and the 5 mounted out.

 

The least complicated is probably going to be what I suggested first, which is with both in the array. The wikis you linked are basically correct except the UI has changed somewhat for v6. New Config is on the Tools page. You will have to reassign all your drives. Make sure you don't assign a data disk to the parity slot!

 

[FrozenGamer]

Thanks Trurl:  So i can move the data from 6 to 5 with both in the array using MC, stop array, then new config ensuring that the the parity disk is kept as parity disk.  Should it all shares etc remain after new config? 

I apologize for repeating questions, but its 80 tb of data, and its a bit scary.

 

 

 

 

[tdallen]

3- copy entire contents of 6 tb drive to empty 5tb drive with MC, stop array and pull out 6tb. (would i not move versus copy?)

I strongly prefer to copy, that way you still have a complete copy of your data on the 6TB if you experience a failure during the operation.  It's easy enough to clear off 6TB after the copy to the 5TB is complete.

[trurl]

Thanks Trurl:  So i can move the data from 6 to 5 with both in the array using MC, stop array, then new config ensuring that the the parity disk is kept as parity disk.  Should it all shares etc remain after new config? 

I apologize for repeating questions, but its 80 tb of data, and its a bit scary.

Your data and shares will be fine. If you wind up changing the order of any disks and you have any shares assigned to specific disk numbers then you will need to update those settings.

 

It's really not that scary if you understand a few basic things about unRAID. Each disk is an independent filesystem so any file exists completely on a single disk. Even if you remove a disk nothing happens to the files on the other disks. The user shares are just the aggregate of all top level folders with the same name, so you can't really get rid of a share anyway unless you intentionally delete all of its data.

[FrozenGamer]

Thanks! I appreciate the help and patience from all.  Just as a matter of note if another linux newb reads this in the future - i am successfully copying the files with the following command after typing mc in telnet.

cp -r -v - /mnt/disk15/* /mnt/disk18

 

Off for a much procrastinated jog on a balmy 48 degree Fahrenheit day in Alaska. 

 

Question for:Trurl - if the same files exist on both disk 15 and 18 is that transparent to the shares/system?  ie - during the 24 hours or so it will take to copy there will be duplicate files, but not visible as duplicates in the share?

[trurl]

Thanks! I appreciate the help and patience from all.  Just as a matter of note if another linux newb reads this in the future - i am successfully copying the files with the following command after typing mc in telnet.

cp -r -v - /mnt/disk15/* /mnt/disk18

 

Off for a much procrastinated jog on a balmy 48 degree Fahrenheit day in Alaska. 

 

Question for:Trurl - if the same files exist on both disk 15 and 18 is that transparent to the shares/system?  ie - during the 24 hours or so it will take to copy there will be duplicate files, but not visible as duplicates in the share?

While it is possible to use command line commands from within mc, the main point of mc is that you can do file management without having to type command line commands. You could have done the copy without typing the cp command, or you could have done the cp command without typing mc.

 

As for duplicates, I think someone recently said it always uses the lowest numbered disk but I haven't tested that.

[FrozenGamer]

OK, looks like i am back online and as far as i can tell the new config worked, however things were very very weird for a bit and it took a few reboots to work.  I had to unplug machine, because the array wouldn't stop.  When i got to the monitor of machine it was showing some errors, which i didn't take time to write down.  (oops)  - I restarted and got access to the webgui and started array.  It was on starting dockers and then the entire network in the house went down, all computers were inaccessable and games got disconnected etc.  I unplugged the network cable off my server and it went away, at that time it had the following error "Tower: login:XFS (md1): metadata I/O error: block 0x74702fd8 ("xfs_trans_read_lf_map") error 5 numblks 16

 

I unplugged it again, rebooted and started array via gui and it all appears to be working fine.  Parity is rebuilding and dockers up and running.

I am seeing the following new smart error though on drive 7

 

188 Command timeout 0x0032 100 099 000 Old age Always Never 4295032833