SMB Lockups When Disks Get Less Than 20% Full Considered "Normal"

[bsim]

I used to have a problem a long time ago, and the answer I got was that I had to keep the disks in the array above 10-20% to keep the SMB service from locking up intermittently. We went through almost every possible step to make sure it was the problem from swapping NICs, attempting safe mode, disabling memory above 4gb.

Finally, I did add quite a bit more new drives into the array, and 95% of my problem went away. But now, I've come to the same issues after dropping to around 15% free on a few of my drives (most free first storage mode). Can anyone tell me if this is still the case with Unraid 6 and if there are any fixes that have happened in the last year for my problem?

 

(unraid pro 5.0.4)

[RobJ]

I wouldn't call it a 'lockup', but the Reiser file system does pause a very long time when it does extra housekeeping as you approach the end of the disk.  With V6, more and more users are converting from ReiserFS to XFS with good results.  However, I really don't know yet how good XFS is at handling full disks, but I suspect it's much better than ReiserFS is.

[bsim]

Is there any way to change the reiser timeouts or forcing it to do housekeeping in the background?

 

Does unraid support BTRFS and any idea if XFS is better than BTRFS?

[RobJ]

Is there any way to change the reiser timeouts or forcing it to do housekeeping in the background?

Messing with Reiser internals is way over any of our pay grades.  But I don't think it's possible, or desirable.  It's almost certainly trying to reorganize file system structures or disk layout because of a write request you have made, so changing it to be even slower would be counter-productive.  You probably don't want to take that write back (couldn't anyway), and it cannot proceed until the housekeeping is done.

 

Does unraid support BTRFS and any idea if XFS is better than BTRFS?

V6.0 betas support both.  See XFS vs BTRFS....

[bsim]

Do either have the same free space requirements as Reiser?

 

Would it ever be recommended to go to something like ext3/4 or another file system altogether?

[RobJ]

Do either have the same free space requirements as Reiser?

I don't think so.  However, you need a response from a true Linux expert, and I'm not.

[bsim]

From other documentation, it looks like XFS may be around 5%, but I'm still researching...

 

Quoting from my research...

 

"My experience is that 'ext3' requires at least 10% free space to perform decently over time (the 5% default is the absolute minimum that should be done)"

 

"Traditionally, filesystems reserve 5-10% of the total space as a mechanism to prevent fragmentation.  With today's gigantic disks and  arrays, that could be a bit excessive, however setting it to zero is a very bad idea."

"Not if you use xfs, set the "minimum free space" and run "xfs_db -rc    frag ..." every night ;-)"

[trurl]

From other documentation, it looks like XFS may be around 5%, but I'm still researching...

 

"My experience is that 'ext3' requires at least 10% free space to perform decently over time (the 5% default is the absolute minimum that should be done)"

ext3 is not the same as XFS. unRAID only supports RFS, XFS, and btrfs for array and cache drives.

[Traxxus]

Is there any way to change the reiser timeouts or forcing it to do housekeeping in the background?

 

Does unraid support BTRFS and any idea if XFS is better than BTRFS?

 

FWIW, Unraid currently defaults to XFS, and I've decided to stick with it since it will probably have the most support/knowledge available.

 

Consensus seems to be that BTRFS can potentially be great/better if promises about future features are met, so depends on whether you want to rely on that or not.

[bsim]

I understand ext3 is not xfs...just attached the quotes of file systems and their free spaces from around the net.

 

I'm guessing that I will migrate to XFS after 6 goes stable. I have heard of a lot of odd bugs occurring in btrfs and everyone loving xfs until btrfs stables out. Hopefully, there won't be too bad of an upgrade path going to 6.