Make wait-for wait for full startup of vrnetlab containers

[JakeBednard]

FR: Make wait-for wait for full startup of vrnetlab containers

One FR that I have is that wait-for cannot sense the full startup of vrnetlab containers. The Dependency Manager senses these containers as started at the level=debug msg="Container started: \"my-container\"" log message. This does not reflect that the vrnetlab containers utilize a full bootup and load sequence on top of qemu, before becoming ready. This can be problematic in resource constrained environments.

One solution would be to utilize launch.py to set a flag within the container after the bootup and configuration apply has been completed. Then the Dependency Manager would need to watch for the flag-flip before marking the dependency as started. I assume this feature would need to signify it presence in the deploy.clab.yml by creating a new node-type or flag to signal the need to watch the flag-flip.

Thanks for adding the wait-for feature @steiler!

[steiler]

Hi @JakeBednard,
I've realized already, that there might be different kinds of dependencies between nodes. See: #1027

Also tried to quickly implement phases like configured (post-deploy) and even healthy (considering runtime health checks).
However, containerlab today runs in parallel for pre-deployment, create and start phase, but then waits for all threads to complete before starting the configuration phase.
Hence introducing dependencies like when x is configured / health start y will result in a deadlock.
If there is enough request, sure I can digg a little deeper and make this all happen. For now, without real demand that seemed to be too much of an effort.
#1027 (comment)