HDD Capacity Off?

[raven]

Hello and thank you for such a great project!

 

I just finished installing Xpenology with DSM 6.1.5 and 5 x 8TB drives. This is a fresh install. I chose to create an SHR-2/BTRFS volume with those 5 HDDs, which should have netted me 24TB of storage, with two drives of parity for fault tolerance. Now, I know this is not really 24TB as each 8TB drive is really 7.28TiB. But still, I should end up with ~21.84TiB of storage space. My system is showing that the volume I created has capacity of 20.95TiB. Where did I lose that one TiB of space? Overhead or SHR-2? If it is because of any of these, are there any alternatives so that I can recoup that lost space? What if I use just plain RAID6 or change to ext4? I still have not put any data on the drives, so can redo/change my previous selections. Any ideas/explanations are welcome.

 

Thanks!

Edited February 6, 2018 by raven

[Dfds]

DSM will be taking some of that space but it's mostly BTRFS overhead, ext4 would give back some space, not sure how much though. As you haven't put any data on yet you could redo the array as ext4 as a comparison.

[sbv3000]

Have you tried re-enabling SHR and setting up an SHR2 array?

 

[raven]

5 hours ago, Dfds said:

DSM will be taking some of that space but it's mostly BTRFS overhead, ext4 would give back some space, not sure how much though. As you haven't put any data on yet you could redo the array as ext4 as a comparison.

 

I could try that since the is no data. The thing is, I have already done this twice, as the first time I ended up with the 20.95TiB on BTRFS. I thought I had done something wrong, so I went back and checked most of the options and all the one s I checked indicated I would en with ~21.82TiB of storage space.

 

36 minutes ago, sbv3000 said:

Have you tried re-enabling SHR and setting up an SHR2 array?

 

 

I do not follow. Re-enable SHR? If you mean editing synoinfo.conf file using WinSCP then yes, I did that right after installation. I could recheck the file.

 

I can try the ext4 to see what I end up with. Another thing may be to reinstall and not modify the file above and just use RAID6 (with its limitations) in either BTRFS or ext4 and see what I end up with. But I am guessing this is BTRFS overhead and a lot of it. Maybe there will be less overhead with SHR1, however I would like more redundancy.

Edited February 6, 2018 by raven

[Dfds]

38 minutes ago, raven said:

I can try the ext4 to see what I end up with. Another thing may be to reinstall and not modify the file above and just use RAID6 (with its limitations) in either BTRFS or ext4 and see what I end up with. But I am guessing this is BTRFS overhead and a lot of it. Maybe there will be less overhead with SHR1, however I would like more redundancy.

Whether you choose to use SHR1, SHR2 or Raid6 I don't think it will make much difference to the available free space as it's BTRFS that's causing the overhead. If the difference in free space between Ext4 & BTRFS is really that much of an issue you need to decide whether the advantages outweigh the disadvantages.

[IG-88]

in SHRx the "x" is the number of redundant disks so in case of 5 identical disks and "2" its the same as raid6, using SHR1 in that hardware scenario is just raid5, the SHR ist about using disks not matching in size to get the most possible space out of it, afaik its basicly "just" mdadm raid's glued together with LVM to one volume

 

the amount of data for the system and swap partition on every disk taken by DSM is just 2 GB + 2.4 GB so with 5 disks it will be ~22GB that will be missing from having system and swap on every disk (thats the way DSM works)

 

also just comparing a formated empty volume might not give the right picture about overhead/metadata, it might be different when the drive is filled with real data

Edited February 6, 2018 by IG-88

[raven]

30 minutes ago, IG-88 said:

in SHRx the "x" is the number of redundant disks so in case of 5 identical disks and "2" its the same as raid6, using SHR1 in that hardware scenario is just raid5, the SHR ist about using disks not matching in size to get the most possible space out of it, afaik its basicly "just" mdadm raid's glued together with LVM to one volume

 

the amount of data for the system and swap partition on every disk taken by DSM is just 2 GB + 2.4 GB so with 5 disks it will be ~22GB that will be missing from having system and swap on every disk (thats the way DSM works)

 

also just comparing a formated empty volume might not give the right picture about overhead/metadata, it might be different when the drive is filled with real data

Thanks for your response. I used SHR2 in case I needed to add storage in the future, especially with unmatched disk sizes. So, if the DSM only takes that small amount of space, what do you suggest happened to my 1TiB? The BTRFS overhead as mentioned above? I will work on this when I get home.

[IG-88]

55 minutes ago, raven said:

what do you suggest happened to my 1TiB? The BTRFS overhead as mentioned above? I will work on this when I get home.

 

i can tell you the numbers from my btrfs raid6 system

 

11 x 3.64TB (so its 9x that number without the redundancy disks)
raid group capacity 32,71TB

volume capacity 31,4TB (btrfs formated)

 

if you have no data on it yet you can reformat with ext4 and compare the numbers, easy thing to do

[raven]

I decided to stick with BTRFS for its benefits. It seems that this ~4% overhead is standard in these situations. Thank you for your help!