[SOLVED] Cache pool wont balance from RAID1 to RAID0

[sneakybifta]

I'm running appdata in a cache pool, I have 1x 120GB and 1 x 256GB SSDs. I've had them in RAID1, but as that only gives me 120GB of space and try as I might to slim it down, my appdata + docker.img is now pushing 115GB. 

 

I've used CA appdata backup to copy everything into the main array, and I'm keeping a copy backed up on my PC too, so I'm not bothered about losing redundancy for now until I can afford a new SSD.

 

I'm trying to change my RAID1 cache pool to RAID0 which should give me 240GB available with increased speed over BTRFS 'single' mode. However I don't seem to be able to switch to either, having tried both and got the same error.

 

I've tried running this and hitting balance, it initially starts balancing but within a few seconds, the first time I hit refresh, it has stopped and changed back to 'No balance found on /mnt/cache'

-dconvert=raid0 -mconvert=raid1

I've tried running the code below as an alternative but the same thing happens. Running it without -mconvert makes no difference either. Filesytem attached. 

 

-dconvert=single -mconvert=raid1

Label: none  uuid: redacted
	Total devices 2 FS bytes used 108.85GiB
	devid    1 size 111.79GiB used 111.79GiB path /dev/sdc1
	devid    2 size 238.47GiB used 111.79GiB path /dev/sdh1

Data, RAID1: total=110.76GiB, used=108.49GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=1.00GiB, used=368.38MiB
GlobalReserve, single: total=121.42MiB, used=0.00B

No balance found on '/mnt/cache'

 

Edited January 13, 2018 by sneakybifta

[JorgeB]

Please post your diagnostics: Tools -> Diagnostics

[sneakybifta]

plexicity-diagnostics-20180113-1140.zip

[JorgeB]

There's no error in the log but the smallest cache is practically full, move some data to the array and try again, also a good idea to first update to v6.4 which includes a newer kernel with many btrfs fixes/improvements.

[sneakybifta]

Just now, johnnie.black said:

There's no error in the log but the smallest cache is practically full, move some data to the array and try again, also a good idea to first update to v6.4 which includes a newer kernel with many btrfs fixes/improvements.

 

Thanks, much appreciated. Just caught the 6.4 post and server is rebooting post update right now. I'll move the 20GB docker image to the array, try again, then if successful move it back afterwards.

[sneakybifta]

This worked, and I am now using RAID0. However, the total RAID0 size was showing 180GB rather than 240GB the calculator said I should get.

 

I tried a rebalance without any input, and now the RAID0 size is showing as 110GB.

 

How can I expand the RAID0 to the full 240GB available to it?

 

 

 

Data, RAID0: total=108.00GiB, used=106.22GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=1.00GiB, used=359.27MiB
GlobalReserve, single: total=113.69MiB, used=0.00B

 

[JorgeB]

16 minutes ago, sneakybifta said:

However, the total RAID0 size was showing 180GB rather than 240GB the calculator said I should get.

That's expected and a known btrfs limitation, when 2 different sized devices are used in a pool the space is incorrectly reported.

 

If the pool is in raid0 you'll be able to use the 240GB.

[sneakybifta]

1 hour ago, johnnie.black said:

That's expected and a known btrfs limitation, when 2 different sized devices are used in a pool the space is incorrectly reported.

 

If the pool is in raid0 you'll be able to use the 240GB.

 

I expected the free space reported in the main WebGUI to be incorrect as it was for RAID1, but the btrfs filesystem on the cache page at least reported the correct Total / Used space.

 

I've swapped from RAID0 to 'single' before I read your reply, and now my btrfs filesystem is showing as below. The total usuable size of the data partition is now showing as only 107GB.

 

Data, single: total=107.00GiB, used=106.31GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=1.00GiB, used=360.97MiB
GlobalReserve, single: total=115.33MiB, used=0.00B

 

[JorgeB]

2 minutes ago, sneakybifta said:

total=107.00GiB,

This is only the current allocated space, not the total device size, to get that use:

 

btrfs fi usage /mnt/cache

 

[sneakybifta]

1 minute ago, johnnie.black said:

This is only the current allocated space, not the total device size, to get that use:

 

btrfs fi usage /mnt/cache

 

 

Ah awesome, I was worried things were about to start breaking due to hitting the limit. I guess the unallocated space will automatically become allocated as my appdata grows, up to the max shown below?
 

Thanks for your help anyway, I'll mark this as solved

 

Overall:
    Device size:                 350.26GiB
    Device allocated:            109.06GiB
    Device unallocated:          241.20GiB
    Device missing:                  0.00B
    Used:                        107.02GiB
    Free (estimated):            241.88GiB      (min: 121.28GiB)
    Data ratio:                       1.00
    Metadata ratio:                   2.00
    Global reserve:              115.42MiB      (used: 0.00B)

Data,single: Size:107.00GiB, Used:106.32GiB
   /dev/sdh1     107.00GiB

Metadata,RAID1: Size:1.00GiB, Used:361.08MiB
   /dev/sdc1       1.00GiB
   /dev/sdh1       1.00GiB

System,RAID1: Size:32.00MiB, Used:16.00KiB
   /dev/sdc1      32.00MiB
   /dev/sdh1      32.00MiB

Unallocated:
   /dev/sdc1     110.76GiB
   /dev/sdh1     130.44GiB

 

[JorgeB]

6 minutes ago, sneakybifta said:

I guess the unallocated space will automatically become allocated as my appdata grows, up to the max shown below?

Yes, it's allocated as needed.