aicam commented on code in PR #6617: URL: https://github.com/apache/texera/pull/6617#discussion_r3639868420
########## bin/k8s/templates/aws/pod-disruption-budgets/app-service-pdbs.yaml: ########## Review Comment: Done — moved to `bin/k8s/templates/aws/pod-disruption-budgets/app-service-pdbs.yaml`. The `podDisruptionBudgets.enabled` gate (default `false`, `true` in `values-aws.yaml`) is unchanged, so on-prem still renders nothing; I noted in the file header that nothing in the set is actually AWS-specific, so any cluster that consolidates nodes can flip the same flag. Also, per your other message: `shared-editing-server` and `pylsp` are now included too, so the set is 10 budgets. Both are single-replica today, where a PDB is inert (evicting the only pod still satisfies `maxUnavailable: 1`), and it starts protecting them as soon as `replicaCount` is raised. ########## bin/k8s/utils/cu-nodepool.yaml: ########## Review Comment: Done. It is now a chart template, `bin/k8s/templates/aws/computing-unit-nodepool.yaml`, gated behind `computingUnitNodePool.enabled` (default `false`, `true` in `values-aws.yaml`) since it needs the `karpenter.sh/v1` CRD. It became a template rather than just a relocated `kubectl` manifest because of your next three comments — the sizing has to come from `values-aws.yaml`, which only works if Helm renders it. `bin/k8s/utils/` is gone from the PR entirely. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
