Windows 10 SMB copying hangs after ~80%

[Michał Fita]

I observed very strange issue with Windows 10 when I finally bring up my tiny FreeNAS based server after 5 months of struggling with complaints about faulty stuff. But that's other story.

My server has two main volumes:

• quardupled, which is RaidZ2 4x2TB HDD
• striped, which is Striped 2x2TB HDD (for offloading non-crucial data)

My first test was transfer of a few films from my Windows 10 based notebook to the SMB share to one of datapools on striped. Nothing special.

To my dreadful surprise copying hangs at some point, usually between 60% and 80% of the whole copying operation. I don't observe any troubles on the NAS box at that time. Similar trials from Windows 7 seem to work fine.
The copy operation initiated from Total Commander 8.52a on the same Windows 10 machine works fine and I copied my files without any pain. Moreover, data transfers achieved by Total Commander seem to be two to three times faster on average.

My FreeNAS version: FreeNAS-9.10-STABLE-201605021851 (35c85f7)

Does anyone observed similar behaviour of the Windows operating system?

 

[anodos]

Nope. Have lots of W10 deployed with no problems accessing freenas.

 

[ajw107]

Hi
Not too helpful, but I've had a similar problem on all recent versions of windows (it actually exists on any OS, but windows timeouts seem to be lower and so cancel the transfer instead of just stalling for a few minutes). What I've noticed is that on my set it happens about every 2.4GB, which may be linked to the amount of RAM my server has (24GB). It only happens when transferring large files (ie 2.4GB of multiple files doesn't seem to trigger the stall, but if there is a file over 2.4GB, it will trigger it). I've ran speed tests on each of the drives; added a SSD as a L2ARC, then as a ZIL; swapped out most of the hard drives to SAS ones (from SATA); made sure all the hard drives are the same size; ooked through loads of logs; and never found an answer for it. It even happens when you copy a large file by SSHing into freenas and doing a local copy from one dataset to another. However I am able to unzip/rar/whatever a multi-file compression set to a single large file over 2.4GB with no problems (so none of the zip or rar part files are over 2.4GB, but the final uncompressed file is). In the end I've just learned to live with it, and not use my NAS concurrently if one machine is transferring a large file. There is also a setting in Windows you can alter to change the samba timeout, which works, but I can never remember what it is, sorry. But I'm sure you'll find it with google.
I know none of this is a solution, but hope it at least helps a bit with finding a solution.
Alex
SERVER: 24GB EEC RAM, ASROCKS Extreme6 Motherboard, Intel i7-4790K, LSI 9211-i8 HBAs, All 4TB SAS drives grouped in 3's and then used to extend the single volume, shares all CIFS

 

[anodos]

Hi
Not too helpful, but I've had a similar problem on all recent versions of windows (it actually exists on any OS, but windows timeouts seem to be lower and so cancel the transfer instead of just stalling for a few minutes). What I've noticed is that on my set it happens about every 2.4GB, which may be linked to the amount of RAM my server has (24GB). It only happens when transferring large files (ie 2.4GB of multiple files doesn't seem to trigger the stall, but if there is a file over 2.4GB, it will trigger it). I've ran speed tests on each of the drives; added a SSD as a L2ARC, then as a ZIL; swapped out most of the hard drives to SAS ones (from SATA); made sure all the hard drives are the same size; ooked through loads of logs; and never found an answer for it. It even happens when you copy a large file by SSHing into freenas and doing a local copy from one dataset to another. However I am able to unzip/rar/whatever a multi-file compression set to a single large file over 2.4GB with no problems (so none of the zip or rar part files are over 2.4GB, but the final uncompressed file is). In the end I've just learned to live with it, and not use my NAS concurrently if one machine is transferring a large file. There is also a setting in Windows you can alter to change the samba timeout, which works, but I can never remember what it is, sorry. But I'm sure you'll find it with google.
I know none of this is a solution, but hope it at least helps a bit with finding a solution.
Alex
SERVER: 24GB EEC RAM, ASROCKS Extreme6 Motherboard, Intel i7-4790K, LSI 9211-i8 HBAs, All 4TB SAS drives grouped in 3's and then used to extend the single volume, shares all CIFS

I transfer lots of gigabytes of data daily and have not seen this behavior. Samba consistently saturates gigabit with large files with connections from Windows 10 clients. No stalling. Perhaps what you see is caused by some underlying hardware or configuration problem.

 

[Michał Fita]

My server is far from top tier... It's for home use, still fancy, but tiny (all in small ITX case :D).

I complained to Microsoft on the feedback hub and after some updates everything is normal now - so I'm able to copy without hanging.