With the new v1.18 release of Polyaxon CLI, admin users (with access to kubectl) can force kill and clean operations in a specific namespace.
Clean all operations in a namespace
Sometime users trying Polyaxon might teardown an installation without invoking the clean-up hooks. Operations started by Polyaxon are created using a CRD (custom resource definition) with a finalizer that ensures, in normal cases, that all resources related to the operations are collected and archived before releasing the resources. Several users ask us why they are not able to delete the pods started by Polyaxon, these pods are managed and reconciled using our CRD.
With the new polyaxon admin clean-ops command users can force clean all operations in a specific namespace:
polyaxon admin clean-ops --namespace K8S_NAMESPACE --deleteClean specific operations
Polyaxon has an operator that manages the lifecycle of the Kubernetes resources (jobs, services, distributed tasks, …) it starts. In some cases, the operator might fail reconciling an operation, and users who try to delete the pods related to these operations notice that the pods get recreated automatically.
With the new polyaxon admin clean-ops command users can force clean specific operations by UUID in a specific namespace:
polyaxon admin clean-ops --namespace K8S_NAMESPACE --uuids UUID1,UUID2,UUID3 --deleteLearn More about Polyaxon
This blog post just goes over a couple of features that we shipped in v1.18. To learn more about all the features, fixes, and enhancements, please visit the release notes and the short term roadmap.
Polyaxon continues to grow quickly and keeps improving and providing the simplest machine learning abstraction. We hope that these updates will improve your workflows and increase your productivity, and again, thank you for your continued feedback and support.
