[flink-kubernetes-operator] branch main updated: [docs] Update doc to ensure that users keep in mind how rollback feature works

[gy...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

gyfora pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/flink-kubernetes-operator.git


The following commit(s) were added to refs/heads/main by this push:
     new f8debf06 [docs] Update doc to ensure that users keep in mind how rollback feature works
f8debf06 is described below

commit f8debf06c369f3843a242803c8653b5c5cfccd18
Author: Oleksandr Nitavskyi <ol...@datadoghq.com>
AuthorDate: Wed Jul 26 15:40:04 2023 +0200

    [docs] Update doc to ensure that users keep in mind how rollback feature works
---
 docs/content/docs/custom-resource/job-management.md | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/docs/content/docs/custom-resource/job-management.md b/docs/content/docs/custom-resource/job-management.md
index a3f37a89..67664069 100644
--- a/docs/content/docs/custom-resource/job-management.md
+++ b/docs/content/docs/custom-resource/job-management.md
@@ -274,6 +274,9 @@ HA is currently required for the rollback functionality.
 Applications are never rolled back to a previous running state if they were suspended before the upgrade.
 In these cases no rollback will be performed.
 
+Rollbacks exclusively affect the `FlinkDeployment`/`FlinkSession` CRDs.
+When releasing a new version of your Flink application that updates several Kubernetes resources (such as config maps and services), you must ensure backward compatibility. This means that your updates to non-FlinkDeployment/FlinkSession resources must be compatible with both old and new FlinkDeployment/FlinkSession resources.
+
 ### Stability condition
 
 Currently, a new job is marked stable as soon as the operator could observe it running. This allows us to detect obvious errors, but it's not always enough to detect more complex issues.