Bamboo Docker task(daemon) failed to execute "mkdir" command

[Fahd]

Error:

build	11-Aug-2015 15:40:07	Sending build context to Docker daemon 

error	11-Aug-2015 15:40:07	time="2015-08-11T13:40:07Z" level=fatal 
msg="Error: mkdir /mnt/docker/tmp/docker-build261033587: no such file or
 directory" 

simple	11-Aug-2015 15:40:07	Failing task since return code of [/usr/bin/docker
 build --force-rm=true --tag="bamboo/XXXXX-XXXXX:221" 
/home/bamboo/bamboo-agent-home/xml-data/build-dir/XXXX-BUILD-JOB1/application/build/distributions] was 1 while expected 0

This is executed in EC2 instance using Atlassian Ubuntu stock image. Restarting Docker service manually in SSH session seem to fix the problem in that running EC2 instance.

[Michael Van Diest · Accepted solution]

You can actually just do the following in the Instance Startup Script:

$ sudo /etc/init.d/docker restart

Not sure why the docker service doesn't come up properly on it's own. But adding this fixed my issue.

[reiner]

Thank you! I forgot to come back to this issue and add this step to the instance startup script entry to my answer.

[reiner]

By following these steps (all the way at the bottom of the page) in the Bamboo instance:

sudo -i
#stop the docker daemon
service docker stop
#remove the broken devicemapper files
rm -rf /var/lib/docker 
service docker start

[Dan DeMaggio]

Have the same problem. Seems to be in the way they remount /home to be the ephemeral drives. Docker containers see the empty /home that is "underneath" the mounted /home that has bamboo stuff on it. Very annoying.

[reiner]

My Docker task also failed to executed the "mkdir" command.

[Michael Van Diest]

Just started having the same problem. Even after launching a few new ec2s. Is this an ubuntu issue? Did the AMI for ubuntu change recently introducing this issue? Not sure...