Thinking a bit more - is this a bug with fabric8?
Assuming the presence of the config map means the JM will try to load HA, and
without the map it starts a job from savepoint dir.
The config map exists - but the fabric8 configmap check is returning false, so
the operator won't create the deploy
Hi all,
I’m looking for some help understanding a behaviour of the k8s operator when
recovering a JobManager deployment.
Scenario is – ApplicationCluster, with HA enabled and upgrade mode of
savepoint. The job is in failed state – so it’s not in the HA config map.
* If I delete the JM pod
+1 to FlinkDeployment helm chart approach. Here is Wikimedia Foundation's
chart: (it is heavily dependent on other WMF templates, so you probably
can't just use it...but maybe!)
https://gerrit.wikimedia.org/r/plugins/gitiles/operations/deployment-charts/+/refs/heads/master/charts/flink-app/
> he
I’d go for a Helm chart, where I would define FlinkDeployment, with desired pod
templates and all other stuff. That other stuff can be your headless service,
PVCs, CofigMaps,… In the pod template you can define also sidecars and init
containers, should you need them.
Nix.
From: Gyula Fóra
Dat
I have created the example table to read data that follows the schema
defined in example.avro.
> *CREATE TABLE example (*
* fieldA STRING,*
* fieldB INT,*
* arrayField ARRAY*
*) WITH (*
*'connector' = 'kafka',*
*'topic' = 'test_flink',*
*'properties.bootstrap.servers' = '***
Hi Yarden!
You have different possibilities:
- Use a podTemplate for your FlinkDeployments
- Extend and build a custom operator version with your specific requirements
- Use a mutating webhook to automatically create a podTemplate for every
FlinkDeployment
Cheers,
Gyula
On Sun, Dec 22, 2024 at 2
