[codespaces] some initial improvements

[jkleinlercher]

while codespaces for our platform seems to work there are some first improvements needed:

• check who can really create a codespace? goal is that everyone with a github account can use it (so no membership in our organization needed)
• port forwarding improvement. maybe we should use NodePorts so application access is possible as soon as the application is up and running. Also in the "Ports-View" the available ports are not always shown?
• it takes a lot of time to set up the platform stack (between 10 and 20 minutes) and maybe new users don't know what is going on. Can we improve the dev experience their? speed up or explain better what is going on (in docs and in codespace). E.g. Creation Logs are not shown out-of-the-box, you must click a link. Maybe this can be configured?
• unmask some output in creation log (e.g. URL)
• starting a stopped codespace results in a damaged k8d instance ( Error from server (ServiceUnavailable): the server is currently unable to handle the request (get namespaces) ) - could be a K3d problem ( https://github.com/k3d-io/k3d/issues/262 ). Maybe switching to KinD prevents this? Or other workaround with K3d (see issue mentioned above)

  
  kubectl cluster-info

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'. Error from server (ServiceUnavailable): apiserver not ready

[jkleinlercher]

Installing the platform stack in the on-start phase (instead of on-create) could be a better approach to forward ports already during setting up. Also, I experienced that problems in on-create result in a alternative container provided by github. Maybe during on-start this is not the case (which would be better).

[jkleinlercher]

do some tests in https://github.com/suxess-it/sx-cnp-oss/tree/feat/improvements-devcontainer/

[jkleinlercher]

Go through open points

[jkleinlercher]

What would be great:

• when argocd is available (synced) it should be extra printed somewhere (output instead of log) or browser should open it automatically
• User/password for argocd should get printed somewhere immediately when argocd is available

[jkleinlercher]

with $BROWSER we can open the website in GitHub codespaces.

but we should open it as soon as the app is synced

unfortunately I get "Command is only available in WSL or inside a Visual Studio Code terminal." when running browser command in script.

[jkleinlercher]

with "::step::" in script output the line will be displayed in the terminal (see https://github.com/microsoft/vscode-dev-containers/issues/1293#issuecomment-1030930939 )

[jkleinlercher]

current status:

done:

• check who can really create a codespace? --> seems to be everyone, a colleague who is not in this organization can create a codescape in sx-cnp-oss repo
• create NodePorts: done
• starting a stopped codespace results in a damaged k8d instance: after switching to KinD this should work better (at least my initial test showed me that)

todo:

• it takes a lot of time to set up the platform stack. improve devex? there is a possibility of prebuild codespaces (but not in the github free plan), and we can show current step in terminal with https://github.com/suxess-it/sx-cnp-oss/issues/592#issuecomment-2366959614
• show ArgoCD URL, username, password somewhere the devs can easily see (also during creating) ... and without masked data
• test again restarting stopped codespaces

[jkleinlercher]

maybe also letting the user do install it manually helps to have a better devex and get no masked output? see https://dynatrace-oss.github.io/progressiveDelivery-masterclass/ also for a tutorial

[jkleinlercher]

I do not think we need/should make it manually, we just need a better documention like in https://dynatrace-oss.github.io/progressiveDelivery-masterclass/ and explain how to access argocd via port tab and how to get the secret. like in https://dynatrace-oss.github.io/progressiveDelivery-masterclass/3-access-argo/

[jkleinlercher]

Since I saw https://dynatrace-oss.github.io/progressiveDelivery-masterclass/ I think quick win would be to create a better documention like in https://dynatrace-oss.github.io/progressiveDelivery-masterclass/ and just document how to access the tools.

Startup-Time is maybe a problem for later. Showing URL without masked is then also not so important when we have a good documention.

So the last thing we should do in this task:

• [x] * create better documention like in https://dynatrace-oss.github.io/progressiveDelivery-masterclass/
• [x] * test every platform-stack
• [x] * test starting an already created codespace (so starting the devcontainer again)
• [x] * for masked output, I started https://github.com/orgs/community/discussions/139593