jabdoa2
jabdoa2
> So this is now with a static backoff-period. I think this is enough for all workloads. I guess that should be good enough for most cases. It would have...
> > I guess we would have to check if the last backoff passed when an update occurs (maybe with a few secs of slack). If that is the case...
> A global cool-down could break the base expectation that there is a restart for the "last" change of a configmap. In general, yes. However, even the Kubernetes api has...
> In that case you require all Deployments to be restarted, you just want to do it slowly, right? Exactly, kubernetes api is fragile as soon as clusters grow. Guess...
> That sounds like it wants to be solved by wave putting the updates into a queue that is slowly drained (leaky bucket for rate limit?) Basically, we already got...
> Like this? Yeah that is great. Simple code and should be very effective. I guess I would set burst to 100 by default. That way wave will act instantly...
I have never seen any other operators behave this way. Flux/ArgoCD with Helm Operator can overwrite/restore deployments but it is easy to tell it to exclude certain path as this...