Various Flink Deployment States

[Mason Chen <ma...@gmail.com>]

Hi all,

There are quite a few states or statuses for a Flink deployment e.g.
deployment status, job status, job manager status etc. I understand these
are useful to debug an error with deployment since there are multiple
points of failure. However, I want to understand how a user can verify that
a deployment upgrade has been successful.

This requires two checks:
1. Deployment spec was consumed by the operator successfully (can check via
current and last generation id).
2. Deployment running successfully (can check that job state is RUNNING)

However, I found a resource lifecycle state [1] and I'm not sure how that
fits in this flow. Is the intention that this is the singular state that
allows the user to check if a deployment upgrade was successful (the
documentation doesn't make it clear, but the source code does suggest
this)? If so, how does this state satisfy point 1 above?

[1]
https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.4/docs/custom-resource/reference/#lifecycleState

Best,
Mason

Re: Various Flink Deployment States

[Mason Chen <ma...@gmail.com>]

Also, I noticed that
https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.4/docs/custom-resource/reference/#jobstatus
reflects the job status provided Flink. Can we use the `JobStatus` enum
from Flink instead of just a String (along the same idea to make using
these statuses a little clearer)?

On Thu, Mar 2, 2023 at 3:10 PM Mason Chen <ma...@gmail.com> wrote:

> Hi all,
>
> There are quite a few states or statuses for a Flink deployment e.g.
> deployment status, job status, job manager status etc. I understand these
> are useful to debug an error with deployment since there are multiple
> points of failure. However, I want to understand how a user can verify that
> a deployment upgrade has been successful.
>
> This requires two checks:
> 1. Deployment spec was consumed by the operator successfully (can check
> via current and last generation id).
> 2. Deployment running successfully (can check that job state is RUNNING)
>
> However, I found a resource lifecycle state [1] and I'm not sure how that
> fits in this flow. Is the intention that this is the singular state that
> allows the user to check if a deployment upgrade was successful (the
> documentation doesn't make it clear, but the source code does suggest
> this)? If so, how does this state satisfy point 1 above?
>
> [1]
> https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.4/docs/custom-resource/reference/#lifecycleState
>
> Best,
> Mason
>