We're in a situation where we run a bunch of kubernetes cronjobs, which will result in 20-30 completed pods lingering around for a while, on various nodes.
This is a bit problematic for the current version of k8s-spot-rescheduler, as it's blocked by said pods.
Upgrading to the latest version of autoscaler, seems to resolve the issue. (related code)
What is the current upgrade path? Upgrading the autoscaler dependency seems to also requiring upgrade of all the other kubernetes dependencies, and that might require a smaller refactor in the code as it is today.
We're in a situation where we run a bunch of kubernetes cronjobs, which will result in 20-30 completed pods lingering around for a while, on various nodes.
This is a bit problematic for the current version of k8s-spot-rescheduler, as it's blocked by said pods.
Upgrading to the latest version of autoscaler, seems to resolve the issue. (related code)
What is the current upgrade path? Upgrading the autoscaler dependency seems to also requiring upgrade of all the other kubernetes dependencies, and that might require a smaller refactor in the code as it is today.