BTRFS


SSD

Recommended Posts

It looks like we're moving away from RFS and towards a more modern file system - namely btrfs.

 

Some of the things I have learned at BTRFS with a little Googling that caught my interest:

 

1 - Btrfs also has "a number of the same design ideas that reiser3/4 had". This is a good thing, I think. Because RFS has done a rather spectacular job in recovering from all sorts of issues - including accidents users have made while trying to recover. It BTRFS can do a similar or better job - it is a great choice IMO.

 

2 - Btrfs maintains checksums and has functions to automatically verify that data was read correctly, and even has features to repair the data with redundancy (i.e, a RAID array it is a part of).

 

I am hopeful that others with some experience with various file systems might explain some of the features they feel are beneficial and why so that the community can better understand the benefits, and if any, drawbacks.

 

I am also specifically curious if btrfs is a good choice for SSD disks.

 

Thanks!

 

Link to comment

Unless it's a file system rap that comes with an inbuilt email feature to notify me of potential failures then it's still not what unRAID needs right now......

 

What allows disparate sized drives in a unRAID setup, RFS or unRAID logic ?  If it's FS dated, does btrfs allow the dame flexibility ?

Link to comment

Unless it's a file system rap that comes with an inbuilt email feature to notify me of potential failures then it's still not what unRAID needs right now......

 

What allows disparate sized drives in a unRAID setup, RFS or unRAID logic ?  If it's FS dated, does btrfs allow the dame flexibility ?

 

I can answer the second question. It is unRAID that allows the drives to be of different sizes.

 

Please let's keep this discussion about btrfs and its features. Complaints belong in the complaints thread, not here.

 

Thanks!

Link to comment

Unless it's a file system rap that comes with an inbuilt email feature to notify me of potential failures then it's still not what unRAID needs right now......

 

What allows disparate sized drives in a unRAID setup, RFS or unRAID logic ?  If it's FS dated, does btrfs allow the dame flexibility ?

 

I can answer the second question. It is unRAID that allows the drives to be of different sizes.

 

Please let's keep this discussion about btrfs and its features. Complaints belong in the complaints thread, not here.

 

Thanks!

 

Thanks.  Though the complaints thread is now the sneak peak thread so I'm struggling to work out where I can have a whinge

Link to comment

well technically speaking BTRFS allows different size drives.it's all depends on how you are using it.

 

what I mean. if unRaid will use btrfs in the same fashion as it uses RFS it does not matter if BTRFS supports different size drives as we will not be using this functionality anyway.

 

now, if you do your research thoroughly, you will know that BTRFS itself  combines features of

LVM and software RAID along with being a CoW file system.

also adds a snapshot functionality to boot.

 

by default BTRFS uses raid1 style storage for metadata and raid0 for data even on single drive setup. meaning when you store a file onto BTRFS volume it will store 2 copy of metadata in different parts of the volume and single copy of data.

you can use raid1 for both metadata and data on single volume, which improve the chances of data recovery do to corruption as you have 2 copy of data and automatic snapshot.

 

now it is possible to create a multi-device spanned volume a-la LVM or RAID-x style using strictly BTRFS, no need the LVM libs or mdam,

even with devices of different size, as BTRFS implements RAID in chunks of (x)MB  so it will try to spread data amongst devices as evenly as possible to  the max size of smallest device in the group. but it is more like the normal raid/lvm style unlike unRaid you can not take the device out of the group and read the data, just like you can not take any device that is part of the LVM volume or Raid array and read the data out of it.

 

 

 

Link to comment

That is what I recall too.  Though SSD was a bonus, I think the main reason was to allow for easy creation of a redundant cache pool.  That way data sitting in the cache isn't at risk to a single drive failure like it is now.  I can see the benefit to others, but I personally don't copy data that is so time sensitive and if it was I'd either run mover more often or use a non-cached share for that important data.  but again, that is me. I can surely see how others might want that level of constant protect + fast writes.

Link to comment

BTRFS brings forth 'trim' support! It will be good for the cache drive (pool).

 

Now if we can use it for array drives, this will be a big plus as it could potentially help us verify our data.

 

What I'm wondering is if it's as recoverable as RFS when someone scribbles on the drive by accidentally pressing that format option.

Link to comment

I can see the benefit to others, but I personally don't copy data that is so time sensitive and if it was I'd either run mover more often or use a non-cached share for that important data.  but again, that is me. I can surely see how others might want that level of constant protect + fast writes.

 

Same here but if I decide to use the virtualization features in anger it would be nice to have some redundancy for the cache drive...  to save time in the event of a failure.

 

 

Link to comment

It looks like we're moving away from RFS and towards a more modern file system - namely btrfs.

 

question, who is "we're"?

are you speaking on behalf of limetech?

 

just trying to understand if this is an announcement or a wish...

 

Take a look at the complaints thread starting on page 20. Jonp has been giving sneak peaks at new features in the upcoming beta, and btrfs (at least for the cache drive initially) is one of them.

Link to comment

It looks like we're moving away from RFS and towards a more modern file system - namely btrfs.

 

question, who is "we're"?

are you speaking on behalf of limetech?

 

just trying to understand if this is an announcement or a wish...

 

Take a look at the complaints thread starting on page 20. Jonp has been giving sneak peaks at new features in the upcoming beta, and btrfs (at least for the cache drive initially) is one of them.

 

thanks will do

lets hope it makes it into the product..

Link to comment

thanks will do

lets hope it makes it into the product..

In the new v6 beta 6 it is a requirement that the cache drive is in btrfs format if you want to start up the docker service.    I think that is a good indication that unless serious issues are found during the beta process it is likely to become a standard feature.

Link to comment

Some questions, no doubt noob questions but alas...

 

Does btrfs support trim for ssd drives ?

Can I use a different label other than cache for the cache drive (to resolve an issue with sickbeard where it ignores all paths with the word cache) ?

Does the cache drive still have to be a single partition, or can I have two or more partitions ?

Link to comment

I understand the improved file protection features available when using btrfs for the cache drive, but how is the performance?

 

Personally, I use the cache drive for temporary storage only, and I can easily write 1TB of temp files to it a day, so I care more about performance than additional layers of protection. The applications I use do CRC checking when reading the temp files on the cache drive, so I have enough protection as it is, but RFS is slower than other filesystems for certain operations, so I also use tmpfs (the RAM drive) for temporary storage for my temporary storage(!)

 

tl;dr, does btrfs support sparse files and how does performance compare with EXT3 or EXT4?

 

The full thread where we discussed RFS performance and sparse files: http://nzbget.net/forum/viewtopic.php?f=3&t=957

 

Will we be able to use EXT3 or EXT4 on the cache drive? If not, why not?

Link to comment

thanks will do

lets hope it makes it into the product..

In the new v6 beta 6 it is a requirement that the cache drive is in btrfs format if you want to start up the docker service.    I think that is a good indication that unless serious issues are found during the beta process it is likely to become a standard feature.

 

I'm hoping it makes it to the data drives as the cache drive is really irrelevant to have such a great file system on it.

 

Link to comment

While I readily admit that I don't understand the whole BTRFS thing.  I read this article on BTRFS a few months ago and found it very interesting.

 

http://arstechnica.com/information-technology/2014/01/bitrot-and-atomic-cows-inside-next-gen-filesystems/

 

I really like the idea of the Snap shots.  Couldn't this in theory replace the parity disc?  Plus if someone accidentally deletes a file you could easily recover it.

 

Chris

Link to comment

While I readily admit that I don't understand the whole BTRFS thing.  I read this article on BTRFS a few months ago and found it very interesting.

 

http://arstechnica.com/information-technology/2014/01/bitrot-and-atomic-cows-inside-next-gen-filesystems/

 

I really like the idea of the Snap shots.  Couldn't this in theory replace the parity disc?  Plus if someone accidentally deletes a file you could easily recover it.

 

Chris

Snapshots are great, but completely different than parity. Parity is used to rebuild the data in the event of partial data loss. snapshots allow you to "roll back" time, but not rebuild data. If the data is gone, the cow snapshot can not rebuild the data. In most cases, the loss would include snapshot data as well. But even will all the snapshot data, only an older version is present. Parity is built in btrfs. It is a conventional raid with data and parity striping.

 

To use your example, a snapshot can be used to restore a file that was accidentally deleted.

But it can not rebuilt a disk that was accidentally shredded. That's what raid parity does, and why it is part of btrfs.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.