search

deislabs / cnab-operator

Experimental CNAB operator for Kubernetes. WIP.
MIT License
7 stars 1 forks source link

Investigate strategy to periodically reconciliate resources #4

Open radu-matei opened 5 years ago

radu-matei commented 5 years ago

Currently, the reconciliation loop ends with requeueing the resource that was just operated on (see this).

We need to investigate if there is another way to periodically reconciliate resources, and if the current solution affects other parts of the controller.

technosophos commented 5 years ago

I believe that @krancour and @jeremyrickard implemented logic like this for OSBA. They might be able to give some pointers.

krancour commented 5 years ago

I think what @technosophos might be referring to is our async angine, which is an implementation of the Redis reliable queue pattern. It also allows us to schedule tasks for future execution. The main value in it is that we don't lose tasks if our process dies. I'm not positive that's what you need here... and will also add a lot of overhead.

Maybe @technosophos was referring to something else that isn't coming to mind for me atm.

technosophos commented 5 years ago

That is what I was referring to