Details
-
Type:
Bug
-
Status: Verified
-
Priority:
Major
-
Resolution: Fixed
-
Fix Version/s: OpenVZ-legacy
-
Component/s: Containers::Kernel
-
Security Level: Public
-
Environment:CentOS 6.7 (Final)
Linux xxxx 2.6.32-042stab112.15 #1 SMP Tue Oct 20 17:22:56 MSK 2015 x86_64 x86_64 x86_64 GNU/Linux
Windows Server 2012 R2 (CIFS share)
Description
>Description of problem:
While trying to copy a file (1GB+) to a mounted CIFS share on a hardware node with an OpenVZ kernel (2.6.32-042stab112.15) enabled, the process ends up hanging.
After switching to the standard CentOS 6.7 kernel (2.6.32-573.12.1.el6.x86_64) the copy of the file is successfully performed.
>How reproducible:
Mount a CIFS share with the following options;
/etc/fstab:
//1.2.3.4/Share /mnt/Share cifs username=XXXX,password=XXXX,gid=48,uid=48,file_mode=0775,dir_mode=0775,_netdev 0 0
>Steps to Reproduce:
1. cp 2GB.bin /mnt/Share
>Actual results:
The process is hanging infinite (it's not ending up as a zombie, I am able to kill the process).
Also I've seen that there is actually portions being copied, for example after 984MB, no extra data was copied, but I have also seen 1004MB, or even much less, like 200MB or something.
>Expected results:
The full file to be copied, and not portions of it which ends up in a hanging process.
I have performed a md5sum on the file that is on the local disk, and the file that is copied to the CIFS share (using a non-OpenVZ kernel) and the copy is exactly the same. This is what I also expect using an OpenVZ kernel.
>Host OS:
CentOS 6.7 (Final)
>Guest OS:
The problem occurs both in (using openvz bind mount with .mount file) and outside the container on the hardware node directly.
>Additional info (see https://openvz.org/Reporting_OpenVZ_problem):
While trying to copy a file (1GB+) to a mounted CIFS share on a hardware node with an OpenVZ kernel (2.6.32-042stab112.15) enabled, the process ends up hanging.
After switching to the standard CentOS 6.7 kernel (2.6.32-573.12.1.el6.x86_64) the copy of the file is successfully performed.
>How reproducible:
Mount a CIFS share with the following options;
/etc/fstab:
//1.2.3.4/Share /mnt/Share cifs username=XXXX,password=XXXX,gid=48,uid=48,file_mode=0775,dir_mode=0775,_netdev 0 0
>Steps to Reproduce:
1. cp 2GB.bin /mnt/Share
>Actual results:
The process is hanging infinite (it's not ending up as a zombie, I am able to kill the process).
Also I've seen that there is actually portions being copied, for example after 984MB, no extra data was copied, but I have also seen 1004MB, or even much less, like 200MB or something.
>Expected results:
The full file to be copied, and not portions of it which ends up in a hanging process.
I have performed a md5sum on the file that is on the local disk, and the file that is copied to the CIFS share (using a non-OpenVZ kernel) and the copy is exactly the same. This is what I also expect using an OpenVZ kernel.
>Host OS:
CentOS 6.7 (Final)
>Guest OS:
The problem occurs both in (using openvz bind mount with .mount file) and outside the container on the hardware node directly.
>Additional info (see https://openvz.org/Reporting_OpenVZ_problem):
Attachments
1.
|
ovz6642 dev subtask |
|
Closed | Vasily Averin |
