147 lines
9.3 KiB
Markdown
147 lines
9.3 KiB
Markdown
# spark-operator
|
|
|
|
 
|
|
|
|
A Helm chart for Spark on Kubernetes operator
|
|
|
|
**Homepage:** <https://github.com/kubeflow/spark-operator>
|
|
|
|
## Introduction
|
|
|
|
This chart bootstraps a [Kubernetes Operator for Apache Spark](https://github.com/kubeflow/spark-operator) deployment using the [Helm](https://helm.sh) package manager.
|
|
|
|
## Prerequisites
|
|
|
|
- Helm >= 3
|
|
- Kubernetes >= 1.16
|
|
|
|
## Previous Helm Chart
|
|
|
|
The previous `spark-operator` Helm chart hosted at [helm/charts](https://github.com/helm/charts) has been moved to this repository in accordance with the [Deprecation timeline](https://github.com/helm/charts#deprecation-timeline). Note that a few things have changed between this version and the old version:
|
|
|
|
- This repository **only** supports Helm chart installations using Helm 3+ since the `apiVersion` on the chart has been marked as `v2`.
|
|
- Previous versions of the Helm chart have not been migrated, and the version has been set to `1.0.0` at the onset. If you are looking for old versions of the chart, it's best to run `helm pull incubator/sparkoperator --version <your-version>` until you are ready to move to this repository's version.
|
|
- Several configuration properties have been changed, carefully review the [values](#values) section below to make sure you're aligned with the new values.
|
|
|
|
## Usage
|
|
|
|
### Add Helm Repo
|
|
|
|
```shell
|
|
helm repo add spark-operator https://kubeflow.github.io/spark-operator
|
|
|
|
helm repo update
|
|
```
|
|
|
|
See [helm repo](https://helm.sh/docs/helm/helm_repo) for command documentation.
|
|
|
|
### Install the chart
|
|
|
|
```shell
|
|
helm install [RELEASE_NAME] spark-operator/spark-operator
|
|
```
|
|
|
|
For example, if you want to create a release with name `spark-operator` in the `default` namespace:
|
|
|
|
```shell
|
|
helm install spark-operator spark-operator/spark-operator
|
|
```
|
|
|
|
Note that `helm` will fail to install if the namespace doesn't exist. Either create the namespace beforehand or pass the `--create-namespace` flag to the `helm install` command.
|
|
|
|
```shell
|
|
helm install spark-operator spark-operator/spark-operator \
|
|
--namespace spark-operator \
|
|
--create-namespace
|
|
```
|
|
|
|
See [helm install](https://helm.sh/docs/helm/helm_install) for command documentation.
|
|
|
|
### Upgrade the chart
|
|
|
|
```shell
|
|
helm upgrade [RELEASE_NAME] spark-operator/spark-operator [flags]
|
|
```
|
|
|
|
See [helm upgrade](https://helm.sh/docs/helm/helm_upgrade) for command documentation.
|
|
|
|
### Uninstall the chart
|
|
|
|
```shell
|
|
helm uninstall [RELEASE_NAME]
|
|
```
|
|
|
|
This removes all the Kubernetes resources associated with the chart and deletes the release, except for the `crds`, those will have to be removed manually.
|
|
|
|
See [helm uninstall](https://helm.sh/docs/helm/helm_uninstall) for command documentation.
|
|
|
|
## Values
|
|
|
|
| Key | Type | Default | Description |
|
|
|-----|------|---------|-------------|
|
|
| affinity | object | `{}` | Affinity for pod assignment |
|
|
| batchScheduler.enable | bool | `false` | Enable batch scheduler for spark jobs scheduling. If enabled, users can specify batch scheduler name in spark application |
|
|
| commonLabels | object | `{}` | Common labels to add to the resources |
|
|
| controllerThreads | int | `10` | Operator concurrency, higher values might increase memory usage |
|
|
| envFrom | list | `[]` | Pod environment variable sources |
|
|
| fullnameOverride | string | `""` | String to override release name |
|
|
| image.pullPolicy | string | `"IfNotPresent"` | Image pull policy |
|
|
| image.repository | string | `"docker.io/kubeflow/spark-operator"` | Image repository |
|
|
| image.tag | string | `""` | if set, override the image tag whose default is the chart appVersion. |
|
|
| imagePullSecrets | list | `[]` | Image pull secrets |
|
|
| ingressUrlFormat | string | `""` | Ingress URL format. Requires the UI service to be enabled by setting `uiService.enable` to true. |
|
|
| istio.enabled | bool | `false` | When using `istio`, spark jobs need to run without a sidecar to properly terminate |
|
|
| labelSelectorFilter | string | `""` | A comma-separated list of key=value, or key labels to filter resources during watch and list based on the specified labels. |
|
|
| leaderElection.lockName | string | `"spark-operator-lock"` | Leader election lock name. Ref: https://github.com/kubeflow/spark-operator/blob/master/docs/user-guide.md#enabling-leader-election-for-high-availability. |
|
|
| leaderElection.lockNamespace | string | `""` | Optionally store the lock in another namespace. Defaults to operator's namespace |
|
|
| logLevel | int | `2` | Set higher levels for more verbose logging |
|
|
| metrics.enable | bool | `true` | Enable prometheus metric scraping |
|
|
| metrics.endpoint | string | `"/metrics"` | Metrics serving endpoint |
|
|
| metrics.port | int | `10254` | Metrics port |
|
|
| metrics.portName | string | `"metrics"` | Metrics port name |
|
|
| metrics.prefix | string | `""` | Metric prefix, will be added to all exported metrics |
|
|
| nameOverride | string | `""` | String to partially override `spark-operator.fullname` template (will maintain the release name) |
|
|
| nodeSelector | object | `{}` | Node labels for pod assignment |
|
|
| podAnnotations | object | `{}` | Additional annotations to add to the pod |
|
|
| podLabels | object | `{}` | Additional labels to add to the pod |
|
|
| podMonitor | object | `{"enable":false,"jobLabel":"spark-operator-podmonitor","labels":{},"podMetricsEndpoint":{"interval":"5s","scheme":"http"}}` | Prometheus pod monitor for operator's pod. |
|
|
| podMonitor.enable | bool | `false` | If enabled, a pod monitor for operator's pod will be submitted. Note that prometheus metrics should be enabled as well. |
|
|
| podMonitor.jobLabel | string | `"spark-operator-podmonitor"` | The label to use to retrieve the job name from |
|
|
| podMonitor.labels | object | `{}` | Pod monitor labels |
|
|
| podMonitor.podMetricsEndpoint | object | `{"interval":"5s","scheme":"http"}` | Prometheus metrics endpoint properties. `metrics.portName` will be used as a port |
|
|
| podSecurityContext | object | `{}` | Pod security context |
|
|
| priorityClassName | string | `""` | A priority class to be used for running spark-operator pod. |
|
|
| rbac.annotations | object | `{}` | Optional annotations for rbac |
|
|
| rbac.create | bool | `false` | **DEPRECATED** use `createRole` and `createClusterRole` |
|
|
| rbac.createClusterRole | bool | `true` | Create and use RBAC `ClusterRole` resources |
|
|
| rbac.createRole | bool | `true` | Create and use RBAC `Role` resources |
|
|
| replicaCount | int | `1` | Desired number of pods, leaderElection will be enabled if this is greater than 1 |
|
|
| resourceQuotaEnforcement.enable | bool | `false` | Whether to enable the ResourceQuota enforcement for SparkApplication resources. Requires the webhook to be enabled by setting `webhook.enable` to true. Ref: https://github.com/kubeflow/spark-operator/blob/master/docs/user-guide.md#enabling-resource-quota-enforcement. |
|
|
| resources | object | `{}` | Pod resource requests and limits Note, that each job submission will spawn a JVM within the Spark Operator Pod using "/usr/local/openjdk-11/bin/java -Xmx128m". Kubernetes may kill these Java processes at will to enforce resource limits. When that happens, you will see the following error: 'failed to run spark-submit for SparkApplication [...]: signal: killed' - when this happens, you may want to increase memory limits. |
|
|
| resyncInterval | int | `30` | Operator resync interval. Note that the operator will respond to events (e.g. create, update) unrelated to this setting |
|
|
| securityContext | object | `{}` | Operator container security context |
|
|
| serviceAccounts.spark.annotations | object | `{}` | Optional annotations for the spark service account |
|
|
| serviceAccounts.spark.create | bool | `true` | Create a service account for spark apps |
|
|
| serviceAccounts.spark.name | string | `""` | Optional name for the spark service account |
|
|
| serviceAccounts.sparkoperator.annotations | object | `{}` | Optional annotations for the operator service account |
|
|
| serviceAccounts.sparkoperator.create | bool | `true` | Create a service account for the operator |
|
|
| serviceAccounts.sparkoperator.name | string | `""` | Optional name for the operator service account |
|
|
| sidecars | list | `[]` | Sidecar containers |
|
|
| sparkJobNamespaces | list | `[""]` | List of namespaces where to run spark jobs |
|
|
| tolerations | list | `[]` | List of node taints to tolerate |
|
|
| uiService.enable | bool | `true` | Enable UI service creation for Spark application |
|
|
| volumeMounts | list | `[]` | |
|
|
| volumes | list | `[]` | |
|
|
| webhook.enable | bool | `false` | Enable webhook server |
|
|
| webhook.namespaceSelector | string | `""` | The webhook server will only operate on namespaces with this label, specified in the form key1=value1,key2=value2. Empty string (default) will operate on all namespaces |
|
|
| webhook.objectSelector | string | `""` | The webhook will only operate on resources with this label/s, specified in the form key1=value1,key2=value2, OR key in (value1,value2). Empty string (default) will operate on all objects |
|
|
| webhook.port | int | `8080` | Webhook service port |
|
|
| webhook.portName | string | `"webhook"` | Webhook container port name and service target port name |
|
|
| webhook.timeout | int | `30` | The annotations applied to init job, required to restore certs deleted by the cleanup job during upgrade |
|
|
|
|
## Maintainers
|
|
|
|
| Name | Email | Url |
|
|
| ---- | ------ | --- |
|
|
| yuchaoran2011 | <yuchaoran2011@gmail.com> | |
|