feat: manage PodDisruptionBudget for SparkApplication driver and executor

[a7i]

Purpose of this PR

Provide the ability to create PodDisruptionBudget per Spark Application

Proposed changes:

• Expose PodDisruptionBudgetSpec for driver definition
• • Expose PodDisruptionBudgetSpec for executor definition

Change Category

• Bugfix (non-breaking change which fixes an issue)
• Feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that could affect existing functionality)
• Documentation update

Rationale

Our spark pipelines cannot be interrupted and during node drain, we want to prevent eviction of executor and driver pods. Once the pipeline is complete, then the node can be drained. This is natively supported via PodDisruptionBudget with maxUnavailable: 0

Checklist

• I have conducted a self-review of my own code.
• I have updated documentation accordingly.
• I have added tests that prove my changes are effective or that my feature works.
• Existing unit tests pass locally with my changes.

Additional Notes

[google-oss-prow]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign yuchaoran2011 for approval. For more information see the Kubernetes Code Review Process.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[missedone]

@a7i , I'm wondering if we can use the new feature of the pod template to specify the PDB per PR #2141

[a7i]

@a7i , I'm wondering if we can use the new feature of the pod template to specify the PDB per PR #2141

@missedone looks like a useful PR! How would pod template control PDB definition? Are you suggesting to implement a single PDB that prevents a common pod label from being evicted?

[missedone]

Ah right, it’s PDB which need a specific configuration item for it. My brain was stuck :(

[jacobsalway]

Happy to take a look at the PR as well because this may be a useful feature, but if this is specifically for node draining, you could add annotations to prevent eviction: karpenter.sh/do-not-disrupt: "true" for Karpenter and "cluster-autoscaler.kubernetes.io/safe-to-evict": "false" for cluster-autoscaler.

[a7i]

"cluster-autoscaler.kubernetes.io/safe-to-evict"

Thanks! The context here is nodegroup upgrades, so we drain the nodes from the old nodegroup. so karpenter or cluster-autoscaler don't come into play.

[Cian911]

This would be a great addition - thanks @a7i

[jacobsalway]

"cluster-autoscaler.kubernetes.io/safe-to-evict"

Thanks! The context here is nodegroup upgrades, so we drain the nodes from the old nodegroup. so karpenter or cluster-autoscaler don't come into play.

Ah, so this is a user initiated drain and not one done automatically by either node provisioner e.g. Karpenter drift detection or node consolidation. In that case definitely understand your issue with needing to provision PDBs.

Will review this tonight.