muvon / yoda Goto Github PK

Sometimes you need control start / restart flow of container. For example, restart each one by one not once on deploy. Allow this possibility with configurable Startfile config

When you init yoda in current project all files goes to docker folder and its hardcoded. Make it customizable and defaults to 'yoda'.

Add colors for deploy statuses displayed in terminal

PATH=\$PATH:~/.yoda ENV=$env GIT_BRANCH=$git_branch REVISION=$rev $custom_args yoda start
Its possible to simplify this call when you switching to selected revision or branch GIT_BRANCH and REVISION could be found automatic using .yodarc definition

Write formula for brew package manager

Sometimes its buggy when using deploy command. It displays with broken new lines without refreshing old ones. Its better to replace escaping characters with tput command to fix cursor position and refresh screen on change.

We should validate flow in Startfile before anything else cuz we can use undefined containers names there and affect starting

You can start one container using yoda start container.0 now. But its better way to start all types of containers for current environment just using its name. For example yoda start container that will start all extra containers with same type too.

Now if we have all containers described in flow of Startfile we also do docker-compose up for rest of containers, so it runs all containers at once with message like 'up to date'.

When run yoda and [container...] it creates container structure but dont add it to Envfile.

When you deploy same node with different projects same time it can be a problem while each project uses own yoda version. It needs to change that behavior using installation of specified version with yoda and use this bin file on each deploy

In start.sh we using ${ENV%%.*}. Its unclear because there is no information in documentation. Think about it and make it more clear that we can have environment and subenv for changing services and number of its services

When you press enter after deploy its rerender only since Elapsed block. Rerender whole screen is better

Make it possible to use shortcut for image name in container.yml. We can take real image name from Buildfile to siimplify

Sometimes you have do magic thing to prepare node. Add prepare command that allow to override it and do things on first node init

Make README file more informative. Describe all features.

Sometimes you need to rebuild Dockerfile without using any cache, not like --rebuild. think about adding this parameter or change rebuild behaviour

Dont start docker build command if that image already was built.

When you init one project with custom Yoda version and Yoda is updated. You could have conflicts.
So its good idea to do it more clever. So when you run Yoda inside current project you just check project version and switch yoda git path to this tag and fork again with right version.

When you deploy and deploy fails you dont know where is to find log. Add hint with log path to CLI command output

You can start all services using yoda start. But you cant stop it and remove all services that was created for current project.

Environment can have namespace followed after dot, for example: production.master to allow make different nodes with different bunch of services running on it.

Before installation check if docker and docker-compose have supported versions

Echo errors from executing scripts to stderr instead of stdout.

Use compose file 2.1 and docker-compose 1.9+

If you create custom script to wrap some command flow you will go into recursive call when calling original command inside it. Need to fix it.

Make --force argument passed to build command rebuild image not depended on if it exists on not.

Yoda builds now every image on node also if this image will not be used. Build only images that will be used on current node with selected containers to run as services.

Now there is env.sh only for all environments. Think about customization it with using extra environment files like env.dev.sh and env.production.sh and so on. Extra files should extend common env.sh file

Its checking docker images -q [image_id] on each iteration of images to build.
Simplify this iteration to get built image once and rebuild only needed without unnecesery checks

Add possibility to update version of Yoda in existing project displaying backwards compatibility as changes plain list on command execution.

For example:
in container.yml

image: some_image_name_that_is_in_my_Byildfile

in another file container.dev.yml

image: some_another_image_name_that_also_in_my_Byildfile

Output from yoda compose command will be like this

image: ${COMPOSE_PROJECT_NAME}/some_image_name_that_is_in_my_Byildfile:${REVISION}
image: some_another_image_name_that_also_in_my_Byildfile

Possible solution:

Redefining can be done by passing multiple compose files for each docker-compose commandс

And this solution has some cool feature like deep-merge for incoming compose files:

compose.first.yml

environment:
  FOO: foo
  BAR: bar

compose.second.yml

environment:
  BAR: baz

docker-compose -f compose.first.yml -f compose.second.yml ... will use this file

environment:
  FOO: foo
  BAR: baz

Display status of deploy process in realtime in deploy command.

Add --force flag to deploy and start command

While upgrading from 1.0 to 1.1 and if it needs container to be removed docker compose hangs on removing orphan container stage.

Now there is only one YML template for container. But what if in development you need to add volumes of source code and dont need it in production?

BASH version check executes in yoda main file now. Its better to move it into Makefile.

Sometimes on deploy error "wait: pid 15757 is not a child of this shell" occurs

Its better to make hostname immutable like container_name in container.yml. Because it has all parts HOSTNAME prefix and container name

On compose stage comments starting with # replace with sequence and break syntax of docker-compose config

When you do first deploy on host its good idea to automate yoda installation on node.

To rollback now you have to know previous commit that were deployed. Need to make it simplier.

Describe each command flow of Yoda in details. For example what exactly happens when deploy command called.

Add or remove container build row in Buildfile on adding or removing container from project.

Add possibility to pass regexp values to --host parameter on yoda deploy command. It will allow to deploy on bunch of servers by custom pattern.

Lock file before run custom script on build command.

When you create image inside project that inherti previous image created you can get bug when you cant build it cuz wrong sequence of building in that case

Now if version does not match you get big bad error message. Make it better to understand for humans.