fdocr / activek8s Goto Github PK

It might be useful to include a "describe pod" task. This will help with debugging and monitoring of the current state of deployments.

The build_deployment task works with simple gsub calls to interpolate image NAMESPACE & IMAGE_TAG.

A proper templating engine might give us a more robust solution, by a lot. This comes to mind considering to manage if / else logic for different NAMESPACES, for example consider the following use case:

Deploy using the autoscaling feature in a Deployment YAML config with up to 5 replicas on a service on 'production' but make use of single replica deployments on staging/development/etv

.ak8s.yml spec

Most of the time, when a port forwarding has been idle for a while the connection dies. I've seen this happen after maybe a minute or two.

Possible solutions are:

• Make use of the healthcheck option in the deployment YAML config to check for a healthy connection. If the healthcheck is dead we can kill and restart the port forward command
• Timeout after a certain time and re-create the port forward command. This is a more rudimentary solution, but easier of course and maybe a fallback for when the YAML specification lacks a healthcheck for the service.

ak8s:init