Allowed Pod disruptions

You can configure the permitted Pod disruptions for Airflow nodes as described in Allowed Pod disruptions.

Unless you configure something else or disable our PodDisruptionBudgets (PDBs), we write the following PDBs:

Schedulers

We only allow a single scheduler to be offline at any given time, regardless of the number of replicas or roleGroups.

Webservers

We only allow a single webserver to be offline at any given time, regardless of the number of replicas or roleGroups.

Executors

  • In the case of Celery executors, we only allow a single executor to be offline at any given time, regardless of the number of replicas or roleGroups.

  • In the case of Kubernetes executors, we don’t deploy any PDB, as it’s Airflows responsibility to take care of the executor Pods.