Review CronJob start deadlines

Description

CronJob starting_deadline_seconds limits how late a job may start after missing its scheduled time. If omitted, this start-delay limit does not apply and a delayed job may run later. Not every missed run is guaranteed to be replayed.

This is not a maximum runtime for an active Job. Jobs that tolerate late execution may omit it; time-sensitive jobs should have a deliberate allowed delay.

Potential impact

A long-delayed job can inappropriately affect current data or state.
An overly short deadline can skip runs even during ordinary controller delays.

Remediation

Where late execution must be limited, set starting_deadline_seconds to the acceptable delay. Account for the controller’s check interval and operational delays.
Make jobs safe to run more than once and review concurrency_policy and failure retries. Configure the Job’s active_deadline_seconds separately if execution time must be limited.

Examples

These existing examples use kubernetes_cron_job syntax. Use kubernetes_cron_job_v1 for configurations using the current API. The after value of 10 seconds is a short illustration, not a suitable default for every job.

Before

hcl

resource "kubernetes_cron_job" "example" {
  metadata {
    name = "demo"
  }

  spec {
    schedule = "1 0 * * *"

    job_template {
      metadata {}
      spec {
        template {
          metadata {}
          spec {
            container {
              name    = "hello"
              image   = "busybox"
              command = ["/bin/sh", "-c", "date; echo Hello from the Kubernetes cluster"]
            }
          }
        }
      }
    }
  }
}

After

hcl

resource "kubernetes_cron_job" "example" {
  metadata {
    name = "demo"
  }

  spec {
    schedule                  = "1 0 * * *"
    starting_deadline_seconds = 10

    job_template {
      metadata {}
      spec {
        template {
          metadata {}
          spec {
            container {
              name    = "hello"
              image   = "busybox"
              command = ["/bin/sh", "-c", "date; echo Hello from the Kubernetes cluster"]
            }
          }
        }
      }
    }
  }
}

Explanation:

Before: No explicit start-delay deadline is set. Review whether late execution is acceptable.
After: A start deadline of 10 seconds after the scheduled time is set. It does not terminate a running Job after 10 seconds.