flink-on-k8s-operator

flink-on-k8s-operator copied to clipboard

Flink Cluster installation is failing with error failed calling webhook

2

Kept the flink operator and flink cluster running together for more than 35 days. I am using helm for installing both flink operator and flink session cluster. When I uninstalled...

vinaykw

cert-job POD failing while installing HELM charts

4

`NAME READY STATUS RESTARTS AGE cert-job-ld89n 0/1 Error 0 12m` `+ kubectl create secret generic webhook-server-cert --from-file=tls.key=/tmp/tmp.aMgt0HWzSq/server-key.pem --from-file=tls.crt=/tmp/tmp.aMgt0HWzSq/server-cert.pem --dry-run -o yaml Error from server (Forbidden): error when retrieving current configuration...

mishra157

upgrade client-go version to 0.21.2

2

Upgrade sigs.k8s.io/controller-runtime from 0.6.0 to 0.9.2, because 0.6.0 is too old to be compatible with many pkgs. Upgrade k8s.io/client-go from 0.18.3 to 0.21.2, because controller-runtime 0.9.2 needs.

johnlanni

The package mod name "googlecloudplatform" of flink-operator is lowercase, but the path name on github is uppercase, which causes the service to need lowercase when referencing this package, but the...

chengjoey

Support cert-manager for admission webhooks

6

In order to automatically renew certificates, cert-manager is widely used in the k8s production environments. I think we should support cert-manager for webhooks.

elanv

exposing a scale resource and way to scale jobs

16

Replaces https://github.com/GoogleCloudPlatform/flink-on-k8s-operator/pull/394 and addresses https://github.com/GoogleCloudPlatform/flink-on-k8s-operator/issues/389 This PR is to break down the complexity of adding autoscaling: It only attempts to add a scaling mechanism that will work for standalone Jobs....

Padarn

Update flink-operator.yaml for helm resource-polic

1

When you set the operator namespace to an existing one, helm delete will recycle the namespace, which is a dangerous operation and will cause all the resources of the namespace...

DevineLiu

Is it possible to support application mode in this operator?

2

@functicons @wangyang0918 I noticed that in Flink 1.12, per job cluster is no longer supported. Flink 1.12 is recommending the application cluster (application mode) instead. In terms of the GCP...

FuyaoLi2017

Hello, We are new to using the operator and have been running a few jobs in production. We don't have HA mode enabled yet and have observed that a few...

yashganti

fix bug for ARG overload

2

executor failed running [/bin/sh -c test -n "$FLINK_VERSION"] ,because in the "FROM" stage, the arg is initialized.

DevineLiu