At the moment I have 4 * 250GB SSD's in a 4 drive BTRFS cache pool.

 

Space is getting scarce and I want to see how I can best increase my capacity in the cache pool.

 

I do not want to add -another- drive and make the pool bigger.

 

The Sandisk Ultra II SSD 960GB looks like an interesting option.

 

How do I go about changing the cache pool ?  With the backup options that we now have in CA I do not need a dual or more drive setup anymore, 1 SSD will suffice.

 

Can I add the new SSD to the cache pool somehow and then "shrink" it to remove the 4 existing ssd's ?

Btrfs is very flexible, you can pretty much do what you want and cache will remain online during the upgrade, only requirement is that the data must always fit, you could for example add the new larger device and then remove the other 4, see the FAQ for the various options.

Btrfs is very flexible, you can pretty much do what you want and cache will remain online during the upgrade, only requirement is that the data must always fit, you could for example add the new larger device and then remove the other 4, see the FAQ for the various options.

 

Great... I want  to make sure I understand.

 

Since I do not care of the redundancy for the cache drives... And I have lack of space.. I could also just convert the whole thing to a jbod and have 1TB available instead of 512GB right ?

 

And the command to do that would be the commend to go for "Single" ?

 

Eg:

 

-dconvert=single -mconvert=raid1

Yes, and since they are the same size you can also use the raid0 profile, same available space with better performance.

They actually are slightly different.. some are 250 and some 255, will that be an issue ?

 

 

Verzonden vanaf mijn iPhone met Tapatalk

There could be a small amount of unused space, if for example 3 are 250GB and 1 is 255GB, 5GB will be unused, if there are 2 of each all space will be used.

That is not a problem whatsoever !

 

I just copied my complete cache drive just be safe and executed the following balance command from the gui:

 

-dconvert=raid0 -mconvert=raid1

 

Mmmm..

 

I did this this morning and it started balancing, but now its finished I do not see any changes...:

 

Label: none  uuid: e01fb837-c696-4b89-9080-e3c019a577a2
Total devices 4 FS bytes used 366.67GiB
devid    1 size 238.47GiB used 228.03GiB path /dev/sdb1
devid    2 size 232.89GiB used 232.70GiB path /dev/sdc1
devid    3 size 232.89GiB used 231.88GiB path /dev/sdd1
devid    4 size 223.57GiB used 188.72GiB path /dev/sde1

 

Data, RAID1: total=240.77GiB, used=206.73GiB
Data, RAID0: total=397.74GiB, used=159.80GiB
System, RAID1: total=32.00MiB, used=112.00KiB
Metadata, RAID1: total=1.00GiB, used=136.31MiB
GlobalReserve, single: total=43.36MiB, used=0.00B

 

I was expecting to see 1TB of space...

 

The command I gave was:

 

-dconvert=raid0 -mconvert=raid1

 

Balance didn't complete, there are two data profiles:

 

Data, RAID1: total=240.77GiB, used=206.73GiB
Data, RAID0: total=397.74GiB, used=159.80GiB

Ok... No idea where that came from... In the FAQ it states that I should seek help on the forum when that is the case :-)

 

I only have a /mnt/cache ..  How can I determine what is in each profile ?  One must be unused.. I cannot imagine anything else..

Try converting again but you may need to free up same space for the conversion to succeed.

 

At the end check btrfs fi df, there should only be one data profile, sometimes it's "normal" to have to run balance more than once.

 

 

I'll give the same command again.. I have 180GB free at the moment.. Sounds like it should be enough ?

Sounds like it should be enough ?

 

It should be enough but these kind of operations can fail if the occupied space is above half, I'll try to find a percentage to add to the FAQ.

 

I know that not long ago I tried to remove a device from a 24 disk raid10 that was about 75% occupied and it failed to move the data to the other devices with an out of space error, when the device I was removing was only 120GB and I had more than 300GB free.

 

If it fails there will be a message in the log with the reason, if it's out of space you'll need to free up some more.

It worked, thanks for the help !!