rails-on-services / ros-cli Goto Github PK

Currently we have development for referring to local development, test points to the K8 cluster, development-test points to circle ci, ...

I think we should clear this configurations to make it closer to what it is:

development => local development
test => circle ci
staging => staging cluster
production => production cluster

After running ros be g service <service_name> for core service (under ros/services) there were some parts that was wrong/missing. Here is the list of all changes that was done afterwards to make it work (on example of Organization service) in a single commit:

rails-on-services/ros@d226448

running ros-native sets the AWS environment variables as they are expected to be set for services that depend on AWS to run.

When running ros be sh <service name> it tries to fetch the AWS caller identity, but because there are env variables set, tries to use those instead. For development this leads to using localstack configuration which is invalid.

I have to unset the AWS keys before i can run any ros-cli commands to run on the kluster.
I think this is a bug

In this run, the process threw an error but the ros cli exited with 0 making CI think that all went well

Including fluentd, kafka, etc.

https://circleci.com/gh/PerxTech/whistler-services/409

if the branch name contains HEAD then means that this has been initiated by a tag.
Get the closest tag git describe --tags --abbrev=0, cut the enable-api- part and use that as the domain for the deploy

when running ros generate service <service name> it tries to start a new application using rails 6.0.0 instead of the 6.0.0.rc2 that we currently use in our services. This then fails because apartment gem is not yet supporting rails 6 creating a breaking dependency.

Either we help on ensuring apartment supports rails 6, or we will have to run rails _6.0.0.rc2_ new <service name> when trying to generate the ros service to ensure that it is initialised with the rc2 version instead of the 6.0.0

During terraform apply (ros be infra apply -v), if user press ctrl-c the ros process exit but terraform apply process still runs in the background

Some running services are displayed as stopped.

We need a mechanism to regularly check for ros-cli updates with convenient command to update it.

I suggest:

• check for updates with any command as post execution action with some interval
• check only if previous check was done more than 24 hours ago
• upon check, print ros command that will do the update of ros-cli, if update needed