Drive full error

[phileeny]

I'm new to unRaid and having moved over from DrivePool, I'm running version 6.3.2, and now my data is on the drives I notice three drives are not been used, SMART says the drives are fine but when I try to move data to any of the three drives via windows 10 it states the drives are full.

Please see the attached picture, Whats wrong and how do I fix it.

[JorgeB]

Post your diagnostics:

 

 

[phileeny]

the whole thing or just part

WDC_WD30EZRX-00D8PB0_WD-WMC4N0957285-20170323-0504.txt

[SSD]

Hmmm ... your smart report looks fine. But did you read down to line 6 in the link that Johnnie posted?

 

If you read there you will see how to pull the diagnostics file.

[phileeny]

yes didn't get a chance to read.

mediaserver-diagnostics-20170323-0504.zip

[JorgeB]

Don't see nothing out of the ordinary on the diags, post the output of both:

 

btrfs fi show /mnt/diskX

 

btrfs fi df /mnt/diskX

 

Replace X with 3, 6 and 10.

 

[phileeny]

I'm not sure how to do that in the webgui but:

Disk 3

Total devices 1 FS bytes used 453.39GiB

devid 1 size 2.73TiB used 457.02GiB path /dev/md3

 

Disk 6

Total devices 1 FS bytes used 1.37TiB

devid 1 size 2.73TiB used 1.38TiB path /dev/md6

 

Disk 10

Total devices 1 FS bytes used 441.62GiB

devid 1 size 2.73TiB used 445.02GiB path /dev/md10

[phileeny]

Disk 3

Data, single: total=456.01GiB, used=452.88GiB

System, single: total=4.00MiB, used=64.00KiB

Matadata, single: total=1.01GiB, used=517.44MiB

GlobalReserve, single: total-512.00MiB, used=0.00B

 

Disk 6

Data, single: total=1.38TiB, used=1.37TiB

System, single: total=4.00MiB, used=176.00KiB

Matadata, single: total=2.01GiB, used=1.50GiB

GlobalReserve, single: total-512.00MiB, used=0.00B

 

Disk 10

Data, single: total=444.01GiB, used=441.11GiB

System, single: total=4.00MiB, used=64.00KiB

Matadata, single: total=1.01GiB, used=520.91MiB

GlobalReserve, single: total-512.92MiB, used=0.00B

[JorgeB]

Everything looks normal also.

 

Can you try a local copy?

 

On the console type mc then on the left pane navigate to to one of your disks, e.g., /mnt/disk1 and on the right pane one of the problem disks, e.g., /mnt/disk3 and try to copy a large file.

[phileeny]

Interesting that seems to work, so some sort of windows 10 issue then.

I have tried to copy files using standard windows copy move and taraCopy

Edited March 23, 2017 by phileeny

[JorgeB]

When copying from Windows were you doing it directly to that disk share, e.g., \\tower\disk3 ?

[phileeny]

no, I set unraid to high-water at first and thats when I saw the disks not been used, then I set unraid to most-free and still not used, when I set unraid to use only disk 3 thats when windows said the disk was full.

I only move files to shares and not disks.

[JorgeB]

OK, try this: Stop the array then go to Settings -> Global Share Settings -> Enable Disk Shares: Set to Yes

 

Start the array again and you will now see all your disks as shares, try to copy directly to one of the affected disks.

 

PS: while enable never copy anything from a disk share to a user share because of the user share copy bug.

[phileeny]

okay i will try that tomorrow and let you know.

[JorgeB]

14 hours ago, johnnie.black said:

Don't see nothing out of the ordinary on the diags

 

There's something wrong in the diags that I didn't check before, df output:

 

Quote

Filesystem      Size  Used Avail Use% Mounted on
rootfs          7.5G  390M  7.1G   6% /
tmpfs           7.5G  300K  7.5G   1% /run
devtmpfs        7.5G     0  7.5G   0% /dev
cgroup_root     7.5G     0  7.5G   0% /sys/fs/cgroup
tmpfs           128M  2.5M  126M   2% /var/log
/dev/sda1        30G  289M   30G   1% /boot
/dev/md1        2.8T  2.4T  349G  88% /mnt/disk1
/dev/md2        2.8T  2.4T  347G  88% /mnt/disk2
/dev/md3        2.8T  454G     0 100% /mnt/disk3
/dev/md4        2.8T  2.2T  625G  78% /mnt/disk4
/dev/md5        2.8T  2.1T  695G  76% /mnt/disk5
/dev/md6        2.8T  1.4T     0 100% /mnt/disk6
/dev/md7        2.8T  2.1T  694G  76% /mnt/disk7
/dev/md8        2.8T  2.1T  695G  76% /mnt/disk8
/dev/md9        2.8T  2.1T  695G  76% /mnt/disk9
/dev/md10       2.8T  443G     0 100% /mnt/disk10
/dev/md11       2.8T  2.1T  689G  76% /mnt/disk11
/dev/md12       2.8T  2.1T  693G  76% /mnt/disk12
/dev/md13       2.8T  2.1T  691G  76% /mnt/disk13
/dev/sdn1       932G   22G  910G   3% /mnt/cache
shfs             36T   24T  6.1T  80% /mnt/user0
shfs             37T   24T  7.0T  78% /mnt/user
/dev/loop0       20G   20M   18G   1% /var/lib/docker
/dev/loop1      1.0G   17M  905M   2% /etc/libvirt

 

As you can see df is reporting the 3 problem disks with 0 free space, no idea why but will try to find out.

[phileeny]

Thanks, I 

On 3/23/2017 at 10:33 PM, johnnie.black said:

OK, try this: Stop the array then go to Settings -> Global Share Settings -> Enable Disk Shares: Set to Yes

 

Start the array again and you will now see all your disks as shares, try to copy directly to one of the affected disks.

 

PS: while enable never copy anything from a disk share to a user share because of the user share copy bug.

I did this and moved a file from windows 10 to Disk 3 with no issues.

very odd.

[JorgeB]

Didn't have much time but did some searching and didn't found any solutions, also it's not a common problem, unlike the reverse, i.e., btrfs file system reports free space but you can't write anymore due to the deice being fully allocated, in fact only found one person running Mageia with the same issue, but for him it was mostly a cosmetic issue, since the system functions normally despite df reporting the disk full.

 

With unRAID looks like the webGUI uses btrfs fi show and correctly reports the free space, but when doing a copy using a user share it looks at df for the available space and since it's not correctly reported it errors out.

 

To fix it you'll probably need to reformat the disk, you can try to resize it first, doubt it will work but it's easy to try, so if you to try it, with the array started type on the console:

 

btrfs filesystem resize 500G /mnt/disk3

Then:

df -h /mnt/disk3

Check if available space changes or if it remains at 0.

 

You can then manually resize the disk to max or it will be automatically resized at next array start.

 

 

 

[phileeny]

47 minutes ago, johnnie.black said:

btrfs filesystem resize 500G /mnt/disk3

I can see what the above does but not the below

 

47 minutes ago, johnnie.black said:

df -h /mnt/disk3

So I typed the above on the console and got this

Size     Used     Avail     Used     Mounted on

2.8T     480G    2.3T     18%      /mnt/disk3

then again

Size     Used     Avail     Used     Mounted on

2.8T     433G    0          100%     /mnt/disk10

So it looks like to me by moving files under MC or via disk share has reset disk3 somehow.

[JorgeB]

Well, that's strange, but all this situation is

 

Try doing the same to the other problem disks to see if they also "normalize".

[phileeny]

Just shared Disk10 and moved a file 1.3mb from windows 10 to disk10, then  df -h /mnt/disk10

 

Size     Used     Avail     Used     Mounted on

2.8T     433G    1.2T     16%      /mnt/disk10

 

Well that worked.

I would like to thank you as I've learned a lot.

[JorgeB]

You're welcome, also learned something, this "solution" should work if the same happens again to another user.

[phileeny]

Hole on there didn't work on disk6

The share disk then move/copy via windows just says disk full, so went to MC and that worked

Size     Used     Avail     Used     Mounted on

2.8T     1.4T    1.4T       51%      /mnt/disk6

 

[JorgeB]

Using mc is probably the best, also noticed disk 10 is still not reporting the correct free space, may need to do it once more later in case it doesn't auto correct.

[phileeny]

probably because I used the windows and not MC in that disk

[SSD]

I would like to thank Johnnie for providing outstanding support to the "unRAID BTRFS pioneers" like phileeny.

 

It is quite likely that BTRFS is the future of file systems, but it is still be in its formative years, and the community is gaining knowledge and experience by which to guide adopters. The fact that BTRFS does not really know how much space remains available on a disk seems to be a recurring theme.

 

For those reading, if your goal is the most rock solid protected array, I would have to suggest sticking with XFS on array disks. If you are choosing to use BTRFS, make sure you follow @johnnie.black!