Jump to content

Reiserfs or XFS?


extremeaudio

Recommended Posts

Posted

Sorry but I did not realize the exact stage at which the filesystem change occurred. I have been a user since v4 and now using 6 stable.

 

So the result of this is that my old server which has been upgraded over the years has a reiserfs file system. And a new server I built since, using one data disk that originally belonged to my older server has 1 disk with reiserfs and the remaining ones with xfs. Is there anything I need to do to get all to xfs since that seems to be the newer way to go?

 

While we are on this topic, I hope the following practice is ok, it has happened a few times:

 

1. Disk 1 on my server has a few movies. Other disks have private stuff.

2. When I want to share those movies with a friend, I just remove the Disk 1 and replace it with a fresh drive. Thus my disk 1 gets rebuilt.

3. Friend in concern pops my disk 1 into his server I build for him along with another fresh disk or 2, set as new config and let the parity sync happen.

 

I don't see any data loss or anything but just wanted to be sure this practice is perfectly fine!

 

Posted

Just keep in mind that during the period starting from removing your disk until it is rebuilt and parity checked good, that any other drive failure will cause data loss. As long as you have good backups of your critical stuff, and can accept the extra risk, I really don't see any issues with your procedure.

Posted

... Is there anything I need to do to get all to xfs since that seems to be the newer way to go?

...

Just be aware that if you change the filesystem of a drive unRAID will format it.
Posted

Just be aware that if you change the filesystem of a drive unRAID will format it.

 

I don't even know how to change the filesystem manually. Are there advantages of having xfs over reiserfs? If yes, I think it would be recommended I start gradually migrating one disk after another to fresh disks and thus create a new xfs server?

Posted

There's no compelling reason to switch older disks to XFS => you certainly want any new disks you add to use XFS, but if you rebuild a disk that has failed (or, in your case, just to make a copy of it), it will still be Reiser ... and that's fine.

 

As for the process you outlined => it's okay, as long as you (a) are doing frequent parity checks so you KNOW you have good parity before doing the rebuild; and (b) realize that any additional drive failure during the rebuild will result in data loss.

 

As already noted, just be sure you have good backups to protect against that possibility.    Note that a forthcoming release will add support for dual parity, which will provide an additional layer of fault tolerance, so a 2nd drive failure during a rebuild won't be a problem.    This is a BIG improvement that many of us are anxiously looking forward to.

 

Posted

If, like many, you simply want all of your disks to be XFS, you can certainly do that -- but be VERY careful and be sure you don't do anything that will result in major data loss from the "user share copy" bug.    [if you don't know about that, search the forum and you'll find plenty.]

 

As I noted, there's really no compelling reason to switch -- if you want to, the SAFEST way to do so is copy all of the data from a Reiser disk to a backup on another system;  then reformat the disk (simply change the file system in the config for that disk and it will reformat); and then copy the data back.

 

My main Media server has 16 disks -- 14 Reiser and 2 XFS -- and I have no plans to migrate the RFS disks to XFS (they're mostly 99+% full anyway, so no new data is written to them).

 

 

 

  • 1 month later...
Posted

They're 3 different file systems.  Reiser is the file system that was used in UnRAID through v5, and is a very reliable system; but has been out-of-favor in the Linux community since its author was convicted of killing his wife and is now in prison.

 

XFS and BTRFS are both based on B-Tree structures, and are both actively supported.  BTRFS is newer and not as mature, but has strong potential with built-in redundancy and protection against bit-rot.

 

I think the most common choices in UnRAID v6 systems are to use XFS for the array and BTRFS for the cache pool (if you have more than one cache drive).

 

There's nothing wrong with mixing the file system types within your array -- as I've noted here and in other threads, if you have full Reiser disks that you don't do much writing to, there's no compelling reason to switch them to XFS.    But for new disks, XFS is a better choice, as it doesn't have the "slow writes to full disks" issue that Reiser does.

 

 

Posted

but has been out-of-favor in the Linux community since its author was convicted of killing his wife and is now in prison.

 

Jesus man!.. I had no clue, feel sorry for the wife!

I don't know if it is enough to where I will boycott his filesystem, but I didn't expect to read that!  :o

Posted

but has been out-of-favor in the Linux community since its author was convicted of killing his wife and is now in prison.

 

Jesus man!.. I had no clue, feel sorry for the wife!

I don't know if it is enough to where I will boycott his filesystem, but I didn't expect to read that!  :o

I think it may be more a case of being out-of-favor due to lack of maintenance rather than an actual boycott.

 

Hans Reiser

Posted

They're 3 different file systems.  Reiser is the file system that was used in UnRAID through v5, and is a very reliable system; but has been out-of-favor in the Linux community since its author was convicted of killing his wife and is now in prison.

 

XFS and BTRFS are both based on B-Tree structures, and are both actively supported.  BTRFS is newer and not as mature, but has strong potential with built-in redundancy and protection against bit-rot.

 

I think the most common choices in UnRAID v6 systems are to use XFS for the array and BTRFS for the cache pool (if you have more than one cache drive).

 

There's nothing wrong with mixing the file system types within your array -- as I've noted here and in other threads, if you have full Reiser disks that you don't do much writing to, there's no compelling reason to switch them to XFS.    But for new disks, XFS is a better choice, as it doesn't have the "slow writes to full disks" issue that Reiser does.

 

Thanks you

Posted

I really like the fact that BTRFS can protect against bit rot, had I know this I would of made my entire array BTRFS instead of XFS, oh well.

if used on a BTRFS data disk it can detect it, but not correct it.  You can also detect it on other file systems using a check summing approach although there it an add-on rather than built-in. The Checksum Suite plugin can provide this as well as recovery based on storing PAR2 recovery data.
Posted

... I think it may be more a case of being out-of-favor due to lack of maintenance rather than an actual boycott.

 

Agree ... it's not actually a boycott => it's simply that since the key developer is no longer available, there hasn't been nearly the level of maintenance that he had always provided, and it's simply fallen out of favor.  It IS, however, a very stable, robust file system -- and in particular the reiserfsck recovery tool is exceptionally good at recovering from corruptions that would also certainly result in lost data on other file systems.

 

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...