Kubernetes node disk full from debug logs

At 20:30, multiple unrelated services on the same Kubernetes node start failing health checks and pods are getting evicted with 'DiskPressure'. The node's root/ephemeral disk is at 100%. A change this afternoon bumped one service's log level to DEBUG, and it's been emitting verbose JSON logs at high volume; container logs aren't being rotated/shipped fast enough. How do you triage, restore the node, and prevent recurrence?