gerkElznik commented on PR #1126:
URL: 
https://github.com/apache/flink-kubernetes-operator/pull/1126#issuecomment-5221257342

   Thanks @Dennis-Mircea, rebased onto main and the conflicts are resolved. The 
restructure had moved every page this PR touched, so the docs changes moved 
with them:
   
   - `deployment/configuration.md` now documents both formats, that 
`config.yaml` takes precedence when both keys are set, and that the resolved 
config is always mounted as `config.yaml`. That replaces the note saying a 
`config.yaml` block is not effectively loadable, since that is the limitation 
this PR removes.
   - `deployment/helm/installation.md` carries the updated values table rows.
   - `internals/startup.md` described the ConfigMap as carrying six data keys 
including `flink-conf.yaml`, and the volume items as "config.yaml (default) or 
flink-conf.yaml". Both are now the single resolved `config.yaml` key, so I 
updated that page and the `GlobalConfiguration` reference in 
`deployment/compatibility.md` along with it.
   - The development guide already refers to `config.yaml` after the 
restructure, so that edit was dropped as redundant.
   - Chinese docs carry the same content.
   
   Re-verified after the rebase: helm-unittest 67/67 and `helm lint` on Helm 
4.2.3, plus `ct lint` on the 3.16.4 image, so both versions the chart CI now 
covers. In a kind cluster with the webhook enabled and an operator image built 
from this branch, the ConfigMap has the single `config.yaml` key (five data 
keys total), the Deployment mounts it as `config.yaml`, and three deployments 
reached RUNNING: `v2_2` with flat and with nested `flinkConfiguration`, and a 
`v1_20` application deployment.
   
   That last one seems worth calling out now that #1152 is in. The operator 
runs on `config.yaml` while writing the v1_20 cluster's config to a 
`flink-conf.yaml` key with a scalar `pipeline.jars`, which is the combination 
that used to crashloop. I also flipped the operator between this chart and 
main's chart under those running jobs, killing the JobManagers after each flip, 
and saw no job disruption, no pod churn, and byte-identical generated configs.
   


-- 
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]

Reply via email to