`docker_utils`

docker_utils.sh is a script to simplify common docker tasks. The tasks are set up as symlinks to docker_utils.sh where the name of symlink determines the task.

Config

The tasks are configured via a local bash script config.sh by setting values for named variables. All *_args variables must be arrays rather than strings for correct parsing of whitespace.

#!/bin/bash

container_name=my_container
build_args=( \
  --no-cache \
  --label maintainer="Warren Moore <warren.moore@spatialbuzz.com>" \
)

Tasks

`build_image.sh`

docker build --rm for the configured image
Image name is set by the config variable container_name
If the environment variable DOCKER_REPO is defined, the image name is set to <DOCKER_REPO>/<container_name>
Extra arguments to the build command can be supplied with the config array build_args e.g. builds_args=(--no-cache)
If command line arguments are supplied, they are added as build options andbuild_args are ignored e.g. ./build_image.sh --no-cache

`start_container.sh`

docker run the configured image
Image and container name is set by the config variable container_name
By default, will run with --rm as a foreground container with the config variable command_args appended to the run command
If the config variable docker_args is defined and non-zero, it is inserted after the run but before the image and command arguments e.g. docker_args=(-v data:/data -p 80:80)
If command line arguments are supplied, will run with -ti as a foreground container and arguments are appended to the run command (config variable command_args is ignored)
If the config variable container_daemon is defined and non-zero is, will run with -d --restart unless-stopped with the config variable daemon_args appended to the run command
- If command line arguments are supplied, these override daemon_args
If the config variable container_network is defined, it is added to the run options as --net <container_network> e.g. container_network=host
If the config variable container_hostname is defined, it is added to the run options as --hostname <container_hostname> otherwise --hostname <container_name> is used (with _ replaced with -)

`stop_container.sh`

For a daemonised container, will run docker stop, docker logs --tail 100 then docker rm -v for the running container
If config variable docker_stop_timeout is set, it is added to the stop options as --time <docker_stop_timeout>

`tail_logs.sh`

docker logs --tail 100 -f <container_name> for the running container

`exec_bash.sh`

docker exec -ti <container_name> bash for the running container

`push_image.sh` and `list_remote_tags.sh`

See Pushing images

`initialise_.sh` and `backup_.sh`

See Volume commands

Link types

If a single Dockerfile can build and run multiple images/containers you can use link types. A link type is specified by adding _<link name> to the task symlink e.g. build_image_v1.sh has the link name v1.

When a task is invoked as a link type, any config variables with the suffix _<link name> will override the standard config variable e.g. build_image_v1.sh would use container_name_v1=my_container_v1 over container_name=my_container.

This applies to all tasks except initialise_*.sh and backup_*.sh.

Pushing images

push_image.sh without command line arguments will show all local images of the current image name
push_image.sh with a command line argument will push the latest build of the current image tagged with the argument provided
If the push_image.sh command line argument is not latest the argument will be used as the image tag
- the local image will be tagged with the argument prior to pushing
- the local git repository will be tagged with the version. NOTE: make sure you have committed all local changes before pushing a versioned image as the script does not check if your git repository has any local changes or nor does it check if the latest image has been rebuilt since any changes have been made locally or remotely (need a CI setup to do this)
- you will be asked if you wish to push the tag to the remote git repository, which is recommended
If you wish to review what tags have been pushed to our ECR repository for the configured image, run list_remote_tags.sh
- This executes and parses aws ecr describe-images --repository-name <*/container_name> so you will need AWS credentials available to run this task e.g. envu ./list_remote_tags.sh

Volume commands

initialise_<volume_id>.sh and backup_<volume_id>.sh are tasks for initialising and backing up named volumes. Like link types, the volumes are referenced by adding the volume id to the symlink. As these commands use the same mechanism as link types, link types cannot be used with these tasks.

The task link volume id is used to dereference the named volume name and mount path within the container.

volume_name_<volume_id> is used to determine the named volume name <container_name>_<volume_name>
volume_dest_<volume_id> is used to determine the mount path within the container

Initialise

initialise_<volume_id>.sh <source path> will run docker create --rm with the named volume <container_name>_<volume_name> mounted on <volume_dest>
The source path is then copied into the container to the location specified by <volume_dest>(via tar -cf - -C <source path> . piped to docker cp)
Once to copy completes, the created container is removed with docker rm -v leaving the seeded named volume

Backup

backup_<volume_id>.sh will create the local backup file backup_<container_name>_<volume_name>.tgz of the destination path (via docker cp piped to gzip to file)

Debug

To trace execution of the task script, the first line docker_utils.sh needs to be set to #!/bin/bash -eux. As config.sh is sourced by docker_utils.sh, changing that will make no difference.

Contact

      @wamonite     - twitter
       \_______.com - web
warren____________/ - email

wamonite/docker_utils

docker_utils