Hi Salva, I think we could have the following options to make Flink application run on a Kubernetes cluster.
1. Local cluster This is what you have in mind. Flink now is really like a common java application, which you could start easily. 2. Standalone cluster on K8s By applying some yaml files, you could create a Flink application cluster. Compared with the local mode, you could have more TaskManagers to get better performance and isolation. Also you could enable the high-availability mode[2] to eliminate the SPOF your Flink application. 3. K8s operator Using K8s operator could make the Flink job submission easier. It could also help with managing the whole lifecycle of the Flink application, including upgrading, restarting, etc. Refer GCP flink-on-k8s-operator[3] for more information. 4. Native Flink on K8s Flink could also natively interact with K8s cluster if the kube config is configured correctly. You need to have a local Flink binary and run a command(e.g. flink run-application ...)[4]. This could also be done in a K8s pod, which is started with Flink official image. [1]. https://ci.apache.org/projects/flink/flink-docs-master/deployment/resource-providers/standalone/kubernetes.html#deploy-application-cluster [2]. https://ci.apache.org/projects/flink/flink-docs-master/deployment/ha/ [3]. https://github.com/GoogleCloudPlatform/flink-on-k8s-operator [4]. https://ci.apache.org/projects/flink/flink-docs-master/deployment/resource-providers/native_kubernetes.html#application-mode Best, Yang Salva Alcántara <salcantara...@gmail.com> 于2021年1月13日周三 上午2:35写道： > I would like to deploy flink on a local cluster built with KIND for the > purposes of e2e testing. The flink app is one of the components running > within the system, which consists of other components (mostly written in > Golang). I was wondering what would be the simplest way for me to deploy > the > flink job for such purposes. Any comments on how to accomplish that would > be > much appreciated. I was thinking on something REALLY simple, using local > execution environment to put everything (flink-related) on a single > container. For instance, locally, I do ` sbt run` for running my job, > something along these lines could do the trick, but that is probably a very > naive approach. > > > > -- > Sent from: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/ >
