Creating a Cache Pool

[expressexcess]

Hi there,

 

I am having a little bit of trouble with the cache pool. I have reduced the number of data drives, and assigned both of my SSDs as cache drives, but the second SSD has a blue ball next to it which the key identifies as "New Device, Not in Pool".

 

 

I have tried stopping the array and reassigning it.

 

When I select the primary cache drive, and then hit "balance" I receive the following output:

 

btrfs balance start /mnt/cache -dconvert=raid1 -mconvert=raid1 2>&1
ERROR: error during balancing '/mnt/cache' - Invalid argument
There may be more info in syslog - try dmesg | tail

 

And this line is produced in my syslog

 

Apr 8 10:16:45 Canteen kernel: BTRFS error (device sdc1): unable to start balance with target data profile 16

 

Can anyone advise me on the proper next steps to get the secondary drive recognized in the pool?

 

Thanks in advance,

Greg

[jonp]

Hi there,

 

I am having a little bit of trouble with the cache pool. I have reduced the number of data drives, and assigned both of my SSDs as cache drives, but the second SSD has a blue ball next to it which the key identifies as "New Device, Not in Pool".

 

 

I have tried stopping the array and reassigning it.

 

When I select the primary cache drive, and then hit "balance" I receive the following output:

 

btrfs balance start /mnt/cache -dconvert=raid1 -mconvert=raid1 2>&1
ERROR: error during balancing '/mnt/cache' - Invalid argument
There may be more info in syslog - try dmesg | tail

 

And this line is produced in my syslog

 

Apr 8 10:16:45 Canteen kernel: BTRFS error (device sdc1): unable to start balance with target data profile 16

 

Can anyone advise me on the proper next steps to get the secondary drive recognized in the pool?

 

Thanks in advance,

Greg

 

Ok, that is definitely odd behavior.  Can you post your syslog here for us to review?  You can download it from the webGui under the tools page (click System Log and then scroll to the bottom and click Download).

[expressexcess]

Ok, that is definitely odd behavior.  Can you post your syslog here for us to review?  You can download it from the webGui under the tools page (click System Log and then scroll to the bottom and click Download).

 

Hi Jon,

 

Thanks for the response. Syslog is attached.

 

Thanks,

Greg

syslog.txt.zip

[jonp]

Apr  8 09:56:35 Canteen emhttp: shcmd (252): set -o pipefail ; /sbin/btrfs device add -f /dev/sdd1 /mnt/cache |& logger
Apr  8 09:56:35 Canteen logger: error checking /dev/sdd1 status: No such file or directory

 

Well, that could have something to do with it ;-)

 

A few questions:

 

Before creating a cache pool, did you start with a single SSD assigned to the cache function and then just added the second one or did you do both of these at the same time?

 

Were either of the SSDs previously used in this or any other system and were they formatted with BTRFS previously?

 

Can you walk me through the steps you took to create the pool (to the best of your memory)?

[expressexcess]

Hi Jon,

 

A few details:

 

- I initially tried to create the pool 2 nights ago on April 6th

- The SSDs were both new in box, and were not formatted as BTRFS previously

 

I was trying to add them both at the same time, but it took me some futzing around to figure out that I needed to reduce the number of data drives. I can't recall if I started the array during this time prior to having both SSDs selected as cache drives. Since then, I have tried stopping the array, and assigning the other disk individually as the sole cache drive which appeared to work.

[jonp]

Hi Jon,

 

A few details:

 

- I initially tried to create the pool 2 nights ago on April 6th

- The SSDs were both new in box, and were not formatted as BTRFS previously

 

I was trying to add them both at the same time, but it took me some futzing around to figure out that I needed to reduce the number of data drives. I can't recall if I started the array during this time prior to having both SSDs selected as cache drives. Since then, I have tried stopping the array, and assigning the other disk individually as the sole cache drive which appeared to work.

Ok, digging deeper into this.

[expressexcess]

Hi all,

 

I removed the disk from the server and tried to repartition it on my Mac (with the intent to let unRAID convert it to BTRFS again). When I started the array, it doesn't appear that it reformatted it, though it recognized it as a BTRFS volume. Oddly, instead of the ~30GB on my cache drive, the cache pool now appeared to be taking up ~130GB.

 

I've unassigned the second cache drive. Using the lsblk command, I see:

 

sdd      8:48   0   477G  0 disk 
??sdd1   8:49   0   200M  0 part 
??sdd2   8:50   0 476.6G  0 part 

 

How can I force a reformat of the disk from the command line or from within unRAID?

[expressexcess]

OK, I think I solved this.

 

I reformatted the drive externally, as MBR and formatted the partition as "free space". I reinstalled the drive back into the server and added a partition using:

sgdisk -g -n 1 /dev/sdd

 

I then assigned the drive and started the array—the drive was recognized by the pool, but the pool was recognizing it as additional free space, rather than being used for redundancy. I selected the primary cache drive and selected "balance".

 

I think this is working properly now!

 

Label: none  uuid: cbcc9fd9-a1dd-459b-9bf3-16a7d332f00f
Total devices 2 FS bytes used 30.20GiB
devid    1 size 476.94GiB used 32.03GiB path /dev/sdc1
devid    2 size 476.94GiB used 32.03GiB path /dev/sdd1

Btrfs v3.18.2

 

Thanks,

Greg

[hgeorges]

Ok, digging deeper into this.

 

Hello;

I've just started exploring v6 and reported somewhere else some personal issues http://lime-technology.com/forum/index.php?topic=40170.0

Meanwhile I'm reading here and there on various items I need to know how to use properly v6.

This posting about setting up a cache pool caught my attention, as I had some issues myself. One should not have to recourse to cli if the Web UI is sufficient. the OP has fixed his problem, but I'm not sure the cache drive setup (a pool) is entirely clear.

 

Is this fully documented somewhere and tested working by different users? 

Here is my server current state:

On the Main page:

Label: none  uuid: a937f5a9-c925-4193-a736-a4c6eef3bb62

Total devices 2 FS bytes used 640.00KiB

devid    1 size 223.57GiB used 2.03GiB path /dev/sdd1

devid    2 size 223.57GiB used 2.03GiB path /dev/sde1

btrfs-progs v4.0

Total: Pool of two devices

 

on the Device Settings page:

Balance status: No balance found on '/mnt/cache'

Is that what it is supposed to be?