starrocks-kubernetes-operator
starrocks-kubernetes-operator copied to clipboard
Kubernetes Operator for StarRocks
## Describe the current behavior StarRocksCluster CRD is saved in a separated folder as shown in [here](https://github.com/StarRocks/starrocks-kubernetes-operator/tree/main/helm-charts/charts/kube-starrocks/charts/operator). In this structure, there's no upgrade support on CRD, since Helm by default...
## Describe the bug This only happens on internal LB (not on external LB). When we uninstall SR cluster, the LB related GCP resource is not cleaned up, making the...
## Describe the current behavior Currently it uses the default behaviour, `Retain` for both `whenDeleted` and `whenScaled`. It leaves PVC un-used when a cluster is deleted or scaled. ## Describe...
## Describe the current behavior Currently when we do a "rolling restart" of the cluster the operator is restarting the pods independent of whether the starrocks cluster is in a...
Currently, operator just simply shutdown BE node when updating the replica number of BE. This will cause data loss. It should be changed to use [DECOMMISSION](https://docs.starrocks.io/zh-cn/latest/administration/Scale_up_down#%E7%BC%A9%E5%AE%B9-be-%E9%9B%86%E7%BE%A4) to scale-in the cluse....
## Describe the current behavior Currently, resource requests are required to be integer strings. Attempting to use milliCPU notation fails. This works: ``` starrocks: cn: cpu: '1' memory: 4Gi replicas:...
CREATE PIPE is awesome help to load data from s3 compatible object storage. It basically processes INSERT SELECT from FILES. An enhancement to it to allow INSERT SELECT form TABLES/VIEWS...
## Describe the bug In shared-nothing mode, when a user executes a scale-in operation (for example, reducing from 6 to 3), and the number of data replicas is 3, then...
## Describe the bug During the scale-down process, because the drop node operation was not executed, a large number of errors appeared in the FE log. ## To Reproduce scale...