I have a brand new unRAID array and have been battling issues with the cache drive. I was trying to transfer a 60gb file to the array today and notice that the transfer speeds weren’t what they should be. I discovered that nothing was being transferred to the cache drive, almost all data was being written to the array. I changed the cache drive settings for the share to “prefer” and files were still going to the array. I change the cache settings to “only” and got an error that the cache drive was full, despite there being 5 240gb SSD’s in RAID 5 and this is literally the first time any data has been transferred to it. When I run 

btrfs fi usage /mnt/cache

This is the output I see:

WARNING: RAID56 detected, not implemented
Overall:
    Device size:                   1.09TiB
    Device allocated:              2.06GiB
    Device unallocated:            1.09TiB
    Device missing:                  0.00B
    Used:                        384.00KiB
    Free (estimated):                0.00B      (min: 8.00EiB)
    Data ratio:                       0.00
    Metadata ratio:                   2.00
    Global reserve:               16.00MiB      (used: 0.00B)

Data,RAID5: Size:24.00GiB, Used:21.51GiB (89.61%)
   /dev/sdb1       6.00GiB
   /dev/sdc1       6.00GiB
   /dev/sdd1       6.00GiB
   /dev/sde1       6.00GiB
   /dev/sdf1       6.00GiB

Metadata,RAID1: Size:1.00GiB, Used:176.00KiB (0.02%)
   /dev/sdd1       1.00GiB
   /dev/sdf1       1.00GiB

System,RAID1: Size:32.00MiB, Used:16.00KiB (0.05%)
   /dev/sdc1      32.00MiB
   /dev/sde1      32.00MiB

Unallocated:
   /dev/sdb1     217.57GiB
   /dev/sdc1     217.54GiB
   /dev/sdd1     216.57GiB
   /dev/sde1     217.54GiB

As you can see, basically all of the data is “unallocated”. What can I do to resolve this?

 

Edit to add diagnostic file.

avenue-stor-01-diagnostics-20200913-1434.zip

Edited September 13, 20205 yr by Smith8154

I am no expert on cache pools, particularly for the RAID aspect, but I a sure that JorgeB will come around to help on that.

It would probably help if you can provide your diagnostics (Tools / Diagnostics) in your next post.

 

 

However, I think your share setup for cache use is incorrect.

If you want the files to go on the cache then on the Array, you need to set the cache to Yes. In a nut shell :

Yes = cache, then array at night

No = direct to the array

Prefer = stay on cache as long as there is enough space

Only = Only stays on cache, never go to the array.

 

For more details, use the help within the unraid interface or ask for details.

I have edited the OP to include the diagnostic file.

 

While this is a new instance of unRAID, I have been using unRAID myself for almost 5 years. I only set the cache settings for the share to “only” to make sure that data written to the share is first sent to the cache as a test. After I get this resolved, I will set the setting back to “yes” so the mover will move the data to the array at night. The “unallocated” part may not even be the issue, that was just my guess.

What directions did you follow to change the RAID level from the stock RAID1?

I didn’t follow any instructions. I just changed the balance to RAID 5.

Try running the balance again.

I’ve run the balance probably 7 or 8 times with no change.

What I meant was balance back to RAID 1, then try going back to RAID 5.

I have tried balancing to RAID 1, RAID 6, and single disk. The value is always lower than 10gb, and when going back to RAID 5 I am now at 8gb.

Try doing a blkdiscard on the /dev/sd? devices that are members of the pool. That will completely erase the devices and allow a fresh format.

Be careful that you properly identify the devices, as all data will be removed.

4 hours ago, Smith8154 said:

As you can see, basically all of the data is “unallocated”.

This is normal with btrfs, it allocates 1GB chunks as need, both for data and metadata, this is not a problem, in fact the opposite can be, if the filesystem is fully allocated when it shouldn't be.

28 minutes ago, JorgeB said:

This is normal with btrfs, it allocates 1GB chunks as need, both for data and metadata, this is not a problem, in fact the opposite can be, if the filesystem is fully allocated when it shouldn't be.

I figured something like that might be the case. But why am I not able to transfer a 60gb file to the array? I should have 960gb of cache space after taking into account the RAID.

 

Edit to add, I can’t even create a folder without getting an error that there is no space available.

Edited September 13, 20205 yr by Smith8154

9 minutes ago, Smith8154 said:

But why am I not able to transfer a 60gb file to the array?

Assuming you mean to cache, because the cache floor is set to 10TB.

15 minutes ago, JorgeB said:

Assuming you mean to cache, because the cache floor is set to 10TB.

I see now. I added one too many 0’s to the minimum free space. It is resolved now. Thank you for pointing out my mistake!