Re: [HTCondor-users] The Docker Universe on execute machine.

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

On 9/17/2018 4:43 AM, Evgeniy Kuznetsov wrote:
> Hi HTCondor community,
> 
> I try to set up docker universe on an execute machine, and follow the 
> "3.16.2" section of HTCondor v8.6.12 manual.
> The Centos 6.10 is installed on the execute host and package docker-io 
> (docker-io.x86_64 1.7.1-2.el6 @epel) as well.
> The user 'condor' added to 'dockerroot' group.
> 
> Service 'docker' is running but the ouput of
> 
> "condor_status -l | grep -i docker"ÂÂ is nothing.
> 
> So that means the execute machine is NOT docker-capable?

Correct. 

> How to correct this ?
> 
> My environment is as follows:
> 
> $CondorVersion: 8.6.12 Aug 06 2018 $
> $CondorPlatform: X86_64-CentOS_6.10 $
> Docker version 1.7.1, build 786b29d/1.7.1
> 
> Evgeny Kuznetsov.


Some quick thoughts on the above:

- The Docker developers only support Docker on Centos 7 and above; Docker itself is not officially supported on Centos 6.  I am guessing you are setting yourself up for pain and suffering trying to run Docker reliably on Centos 6.  Maybe if you updated the kernel...

- What is the 'dockerroot' group?  To give HTCondor permission to run docker commands, you would add user 'condor' to the 'docker' group, i.e.

  # usermod -aG docker condor

- Did you restart HTCondor (or condor_reconfig) after adding user 'condor' to 'docker' group?

- As user root, does the following command work properly?
  
  docker run hello-world

- Does the above docker run command also work as user condor?

- Docker is a very fast moving project; bugs fix releases happen often.  The version of Docker you are truing to use from EPEL (docker.io) is likely significantly older (and thus has more bugs) than the latest version of Docker CE you would find at download.docker.com.   I would suggest using the latest version of Docker (and we will change the instructions in the HTCondor documentation to follow this advice!).  If it helps, here is what I did on my Centos 7 machine to activate HTCondor's docker support on an execute node using the latest free Docker version (run all commands as root):

  systemctl stop condor
  yum install -y yum-utils device-mapper-persistent-data lvm2
  yum-config-manager --add-repo https://download.docker.com/linux/centos/docker-ce.repo
  yum install -y docker-ce
  systemctl start docker
  systemctl enable docker
  usermod -aG docker condor
  systemctl start condor
 
Hope the above helps,
Todd