分布式调度框架。
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

54 KiB

DolphinScheduler

DolphinScheduler is a distributed and easy-to-expand visual DAG workflow scheduling system, dedicated to solving the complex dependencies in data processing, making the scheduling system out of the box for data processing.

Introduction

This chart bootstraps a DolphinScheduler distributed deployment on a Kubernetes cluster using the Helm package manager.

Prerequisites

  • Helm 3.1.0+
  • Kubernetes 1.12+
  • PV provisioner support in the underlying infrastructure

Installing the Chart

To install the chart with the release name dolphinscheduler:

$ git clone https://github.com/apache/incubator-dolphinscheduler.git
$ cd incubator-dolphinscheduler/docker/kubernetes/dolphinscheduler
$ helm repo add bitnami https://charts.bitnami.com/bitnami
$ helm dependency update .
$ helm install dolphinscheduler .

To install the chart with a namespace named test:

$ helm install dolphinscheduler . -n test

Tip: If a namespace named test is used, the option -n test needs to be added to the helm and kubectl command

These commands deploy DolphinScheduler on the Kubernetes cluster in the default configuration. The configuration section lists the parameters that can be configured during installation.

Tip: List all releases using helm list

Access DolphinScheduler UI

If ingress.enabled in values.yaml is set to true, you just access http://${ingress.host}/dolphinscheduler in browser.

Tip: If there is a problem with ingress access, please contact the Kubernetes administrator and refer to the Ingress

Otherwise, you need to execute port-forward command like:

$ kubectl port-forward --address 0.0.0.0 svc/dolphinscheduler-api 12345:12345
$ kubectl port-forward --address 0.0.0.0 -n test svc/dolphinscheduler-api 12345:12345 # with test namespace

Tip: If the error of unable to do port forwarding: socat not found appears, you need to install socat at first

And then access the web: http://192.168.xx.xx:12345/dolphinscheduler

The default username is admin and the default password is dolphinscheduler123

Tip: For quick start in docker, you can create a tenant named ds and associate the user admin with the tenant ds

Uninstalling the Chart

To uninstall/delete the dolphinscheduler deployment:

$ helm uninstall dolphinscheduler

The command removes all the Kubernetes components but PVC's associated with the chart and deletes the release.

To delete the PVC's associated with dolphinscheduler:

$ kubectl delete pvc -l app.kubernetes.io/instance=dolphinscheduler

Note: Deleting the PVC's will delete all data as well. Please be cautious before doing it.

Configuration

The Configuration file is values.yaml, and the following tables lists the configurable parameters of the DolphinScheduler chart and their default values.

Parameter Description Default
timezone World time and date for cities in all time zones Asia/Shanghai
image.repository Docker image repository for the DolphinScheduler apache/dolphinscheduler
image.tag Docker image version for the DolphinScheduler latest
image.pullPolicy Image pull policy. One of Always, Never, IfNotPresent IfNotPresent
image.pullSecret Image pull secret. An optional reference to secret in the same namespace to use for pulling any of the images nil
postgresql.enabled If not exists external PostgreSQL, by default, the DolphinScheduler will use a internal PostgreSQL true
postgresql.postgresqlUsername The username for internal PostgreSQL root
postgresql.postgresqlPassword The password for internal PostgreSQL root
postgresql.postgresqlDatabase The database for internal PostgreSQL dolphinscheduler
postgresql.persistence.enabled Set postgresql.persistence.enabled to true to mount a new volume for internal PostgreSQL false
postgresql.persistence.size PersistentVolumeClaim Size 20Gi
postgresql.persistence.storageClass PostgreSQL data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
externalDatabase.type If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database type will use it postgresql
externalDatabase.driver If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database driver will use it org.postgresql.Driver
externalDatabase.host If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database host will use it localhost
externalDatabase.port If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database port will use it 5432
externalDatabase.username If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database username will use it root
externalDatabase.password If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database password will use it root
externalDatabase.database If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database database will use it dolphinscheduler
externalDatabase.params If exists external PostgreSQL, and set postgresql.enabled value to false. DolphinScheduler's database params will use it characterEncoding=utf8
zookeeper.enabled If not exists external Zookeeper, by default, the DolphinScheduler will use a internal Zookeeper true
zookeeper.fourlwCommandsWhitelist A list of comma separated Four Letter Words commands to use srvr,ruok,wchs,cons
zookeeper.service.port ZooKeeper port 2181
zookeeper.persistence.enabled Set zookeeper.persistence.enabled to true to mount a new volume for internal Zookeeper false
zookeeper.persistence.size PersistentVolumeClaim Size 20Gi
zookeeper.persistence.storageClass Zookeeper data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
zookeeper.zookeeperRoot Specify dolphinscheduler root directory in Zookeeper /dolphinscheduler
externalZookeeper.zookeeperQuorum If exists external Zookeeper, and set zookeeper.enabled value to false. Specify Zookeeper quorum 127.0.0.1:2181
externalZookeeper.zookeeperRoot If exists external Zookeeper, and set zookeeper.enabled value to false. Specify dolphinscheduler root directory in Zookeeper /dolphinscheduler
common.configmap.DOLPHINSCHEDULER_ENV System env path, self configuration, please read values.yaml []
common.configmap.DOLPHINSCHEDULER_DATA_BASEDIR_PATH User data directory path, self configuration, please make sure the directory exists and have read write permissions /tmp/dolphinscheduler
common.configmap.RESOURCE_STORAGE_TYPE Resource storage type: HDFS, S3, NONE HDFS
common.configmap.RESOURCE_UPLOAD_PATH Resource store on HDFS/S3 path, please make sure the directory exists on hdfs and have read write permissions /dolphinscheduler
common.configmap.FS_DEFAULT_FS Resource storage file system like file:///, hdfs://mycluster:8020 or s3a://dolphinscheduler file:///
common.configmap.FS_S3A_ENDPOINT S3 endpoint when common.configmap.RESOURCE_STORAGE_TYPE is set to S3 s3.xxx.amazonaws.com
common.configmap.FS_S3A_ACCESS_KEY S3 access key when common.configmap.RESOURCE_STORAGE_TYPE is set to S3 xxxxxxx
common.configmap.FS_S3A_SECRET_KEY S3 secret key when common.configmap.RESOURCE_STORAGE_TYPE is set to S3 xxxxxxx
common.fsFileResourcePersistence.enabled Set common.fsFileResourcePersistence.enabled to true to mount a new file resource volume for api and worker false
common.fsFileResourcePersistence.accessModes PersistentVolumeClaim Access Modes, must be ReadWriteMany [ReadWriteMany]
common.fsFileResourcePersistence.storageClassName Resource Persistent Volume Storage Class, must support the access mode: ReadWriteMany -
common.fsFileResourcePersistence.storage PersistentVolumeClaim Size 20Gi
master.podManagementPolicy PodManagementPolicy controls how pods are created during initial scale up, when replacing pods on nodes, or when scaling down Parallel
master.replicas Replicas is the desired number of replicas of the given Template 3
master.annotations The annotations for master server {}
master.affinity If specified, the pod's scheduling constraints {}
master.nodeSelector NodeSelector is a selector which must be true for the pod to fit on a node {}
master.tolerations If specified, the pod's tolerations {}
master.resources The resource limit and request config for master server {}
master.configmap.DOLPHINSCHEDULER_OPTS The java options for master server ""
master.configmap.MASTER_EXEC_THREADS Master execute thread number 100
master.configmap.MASTER_EXEC_TASK_NUM Master execute task number in parallel 20
master.configmap.MASTER_HEARTBEAT_INTERVAL Master heartbeat interval 10
master.configmap.MASTER_TASK_COMMIT_RETRYTIMES Master commit task retry times 5
master.configmap.MASTER_TASK_COMMIT_INTERVAL Master commit task interval 1000
master.configmap.MASTER_MAX_CPULOAD_AVG Only less than cpu avg load, master server can work. default value : the number of cpu cores * 2 100
master.configmap.MASTER_RESERVED_MEMORY Only larger than reserved memory, master server can work. default value : physical memory * 1/10, unit is G 0.1
master.configmap.MASTER_LISTEN_PORT Master listen port 5678
master.livenessProbe.enabled Turn on and off liveness probe true
master.livenessProbe.initialDelaySeconds Delay before liveness probe is initiated 30
master.livenessProbe.periodSeconds How often to perform the probe 30
master.livenessProbe.timeoutSeconds When the probe times out 5
master.livenessProbe.failureThreshold Minimum consecutive successes for the probe 3
master.livenessProbe.successThreshold Minimum consecutive failures for the probe 1
master.readinessProbe.enabled Turn on and off readiness probe true
master.readinessProbe.initialDelaySeconds Delay before readiness probe is initiated 30
master.readinessProbe.periodSeconds How often to perform the probe 30
master.readinessProbe.timeoutSeconds When the probe times out 5
master.readinessProbe.failureThreshold Minimum consecutive successes for the probe 3
master.readinessProbe.successThreshold Minimum consecutive failures for the probe 1
master.persistentVolumeClaim.enabled Set master.persistentVolumeClaim.enabled to true to mount a new volume for master false
master.persistentVolumeClaim.accessModes PersistentVolumeClaim Access Modes [ReadWriteOnce]
master.persistentVolumeClaim.storageClassName Master logs data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
master.persistentVolumeClaim.storage PersistentVolumeClaim Size 20Gi
worker.podManagementPolicy PodManagementPolicy controls how pods are created during initial scale up, when replacing pods on nodes, or when scaling down Parallel
worker.replicas Replicas is the desired number of replicas of the given Template 3
worker.annotations The annotations for worker server {}
worker.affinity If specified, the pod's scheduling constraints {}
worker.nodeSelector NodeSelector is a selector which must be true for the pod to fit on a node {}
worker.tolerations If specified, the pod's tolerations {}
worker.resources The resource limit and request config for worker server {}
worker.configmap.DOLPHINSCHEDULER_OPTS The java options for worker server ""
worker.configmap.WORKER_EXEC_THREADS Worker execute thread number 100
worker.configmap.WORKER_HEARTBEAT_INTERVAL Worker heartbeat interval 10
worker.configmap.WORKER_MAX_CPULOAD_AVG Only less than cpu avg load, worker server can work. default value : the number of cpu cores * 2 100
worker.configmap.WORKER_RESERVED_MEMORY Only larger than reserved memory, worker server can work. default value : physical memory * 1/10, unit is G 0.1
worker.configmap.WORKER_LISTEN_PORT Worker listen port 1234
worker.configmap.WORKER_GROUPS Worker groups default
worker.configmap.WORKER_HOST_WEIGHT Worker host weight 100
worker.livenessProbe.enabled Turn on and off liveness probe true
worker.livenessProbe.initialDelaySeconds Delay before liveness probe is initiated 30
worker.livenessProbe.periodSeconds How often to perform the probe 30
worker.livenessProbe.timeoutSeconds When the probe times out 5
worker.livenessProbe.failureThreshold Minimum consecutive successes for the probe 3
worker.livenessProbe.successThreshold Minimum consecutive failures for the probe 1
worker.readinessProbe.enabled Turn on and off readiness probe true
worker.readinessProbe.initialDelaySeconds Delay before readiness probe is initiated 30
worker.readinessProbe.periodSeconds How often to perform the probe 30
worker.readinessProbe.timeoutSeconds When the probe times out 5
worker.readinessProbe.failureThreshold Minimum consecutive successes for the probe 3
worker.readinessProbe.successThreshold Minimum consecutive failures for the probe 1
worker.persistentVolumeClaim.enabled Set worker.persistentVolumeClaim.enabled to true to enable persistentVolumeClaim for worker false
worker.persistentVolumeClaim.dataPersistentVolume.enabled Set worker.persistentVolumeClaim.dataPersistentVolume.enabled to true to mount a data volume for worker false
worker.persistentVolumeClaim.dataPersistentVolume.accessModes PersistentVolumeClaim Access Modes [ReadWriteOnce]
worker.persistentVolumeClaim.dataPersistentVolume.storageClassName Worker data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
worker.persistentVolumeClaim.dataPersistentVolume.storage PersistentVolumeClaim Size 20Gi
worker.persistentVolumeClaim.logsPersistentVolume.enabled Set worker.persistentVolumeClaim.logsPersistentVolume.enabled to true to mount a logs volume for worker false
worker.persistentVolumeClaim.logsPersistentVolume.accessModes PersistentVolumeClaim Access Modes [ReadWriteOnce]
worker.persistentVolumeClaim.logsPersistentVolume.storageClassName Worker logs data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
worker.persistentVolumeClaim.logsPersistentVolume.storage PersistentVolumeClaim Size 20Gi
alert.replicas Replicas is the desired number of replicas of the given Template 1
alert.strategy.type Type of deployment. Can be "Recreate" or "RollingUpdate" RollingUpdate
alert.strategy.rollingUpdate.maxSurge The maximum number of pods that can be scheduled above the desired number of pods 25%
alert.strategy.rollingUpdate.maxUnavailable The maximum number of pods that can be unavailable during the update 25%
alert.annotations The annotations for alert server {}
alert.affinity If specified, the pod's scheduling constraints {}
alert.nodeSelector NodeSelector is a selector which must be true for the pod to fit on a node {}
alert.tolerations If specified, the pod's tolerations {}
alert.resources The resource limit and request config for alert server {}
alert.configmap.DOLPHINSCHEDULER_OPTS The java options for alert server ""
alert.configmap.ALERT_PLUGIN_DIR Alert plugin directory lib/plugin/alert
alert.livenessProbe.enabled Turn on and off liveness probe true
alert.livenessProbe.initialDelaySeconds Delay before liveness probe is initiated 30
alert.livenessProbe.periodSeconds How often to perform the probe 30
alert.livenessProbe.timeoutSeconds When the probe times out 5
alert.livenessProbe.failureThreshold Minimum consecutive successes for the probe 3
alert.livenessProbe.successThreshold Minimum consecutive failures for the probe 1
alert.readinessProbe.enabled Turn on and off readiness probe true
alert.readinessProbe.initialDelaySeconds Delay before readiness probe is initiated 30
alert.readinessProbe.periodSeconds How often to perform the probe 30
alert.readinessProbe.timeoutSeconds When the probe times out 5
alert.readinessProbe.failureThreshold Minimum consecutive successes for the probe 3
alert.readinessProbe.successThreshold Minimum consecutive failures for the probe 1
alert.persistentVolumeClaim.enabled Set alert.persistentVolumeClaim.enabled to true to mount a new volume for alert false
alert.persistentVolumeClaim.accessModes PersistentVolumeClaim Access Modes [ReadWriteOnce]
alert.persistentVolumeClaim.storageClassName Alert logs data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
alert.persistentVolumeClaim.storage PersistentVolumeClaim Size 20Gi
api.replicas Replicas is the desired number of replicas of the given Template 1
api.strategy.type Type of deployment. Can be "Recreate" or "RollingUpdate" RollingUpdate
api.strategy.rollingUpdate.maxSurge The maximum number of pods that can be scheduled above the desired number of pods 25%
api.strategy.rollingUpdate.maxUnavailable The maximum number of pods that can be unavailable during the update 25%
api.annotations The annotations for api server {}
api.affinity If specified, the pod's scheduling constraints {}
api.nodeSelector NodeSelector is a selector which must be true for the pod to fit on a node {}
api.tolerations If specified, the pod's tolerations {}
api.resources The resource limit and request config for api server {}
api.configmap.DOLPHINSCHEDULER_OPTS The java options for api server ""
api.livenessProbe.enabled Turn on and off liveness probe true
api.livenessProbe.initialDelaySeconds Delay before liveness probe is initiated 30
api.livenessProbe.periodSeconds How often to perform the probe 30
api.livenessProbe.timeoutSeconds When the probe times out 5
api.livenessProbe.failureThreshold Minimum consecutive successes for the probe 3
api.livenessProbe.successThreshold Minimum consecutive failures for the probe 1
api.readinessProbe.enabled Turn on and off readiness probe true
api.readinessProbe.initialDelaySeconds Delay before readiness probe is initiated 30
api.readinessProbe.periodSeconds How often to perform the probe 30
api.readinessProbe.timeoutSeconds When the probe times out 5
api.readinessProbe.failureThreshold Minimum consecutive successes for the probe 3
api.readinessProbe.successThreshold Minimum consecutive failures for the probe 1
api.persistentVolumeClaim.enabled Set api.persistentVolumeClaim.enabled to true to mount a new volume for api false
api.persistentVolumeClaim.accessModes PersistentVolumeClaim Access Modes [ReadWriteOnce]
api.persistentVolumeClaim.storageClassName api logs data Persistent Volume Storage Class. If set to "-", storageClassName: "", which disables dynamic provisioning -
api.persistentVolumeClaim.storage PersistentVolumeClaim Size 20Gi
ingress.enabled Enable ingress false
ingress.host Ingress host dolphinscheduler.org
ingress.path Ingress path /dolphinscheduler
ingress.tls.enabled Enable ingress tls false
ingress.tls.secretName Ingress tls secret name dolphinscheduler-tls

FAQ

How to use MySQL as the DolphinScheduler's database instead of PostgreSQL?

Because of the commercial license, we cannot directly use the driver and client of MySQL.

If you want to use MySQL, you can build a new image based on the apache/dolphinscheduler image as follows.

  1. Download the MySQL driver mysql-connector-java-5.1.49.jar (require >=5.1.47)

  2. Create a new Dockerfile to add MySQL driver and client:

FROM apache/dolphinscheduler:latest
COPY mysql-connector-java-5.1.49.jar /opt/dolphinscheduler/lib
RUN apk add --update --no-cache mysql-client
  1. Build a new docker image including MySQL driver and client:
docker build -t apache/dolphinscheduler:mysql .
  1. Push the docker image apache/dolphinscheduler:mysql to a docker registry

  2. Modify image repository and update tag to mysql in values.yaml

  3. Modify postgresql enabled to false

  4. Modify externalDatabase (especially modify host, username and password):

externalDatabase:
  type: "mysql"
  driver: "com.mysql.jdbc.Driver"
  host: "localhost"
  port: "3306"
  username: "root"
  password: "root"
  database: "dolphinscheduler"
  params: "useUnicode=true&characterEncoding=UTF-8"
  1. Run a DolphinScheduler release in Kubernetes (See Installing the Chart)

How to support MySQL datasource in Datasource manage?

Because of the commercial license, we cannot directly use the driver of MySQL.

If you want to add MySQL datasource, you can build a new image based on the apache/dolphinscheduler image as follows.

  1. Download the MySQL driver mysql-connector-java-5.1.49.jar (require >=5.1.47)

  2. Create a new Dockerfile to add MySQL driver:

FROM apache/dolphinscheduler:latest
COPY mysql-connector-java-5.1.49.jar /opt/dolphinscheduler/lib
  1. Build a new docker image including MySQL driver:
docker build -t apache/dolphinscheduler:mysql-driver .
  1. Push the docker image apache/dolphinscheduler:mysql-driver to a docker registry

  2. Modify image repository and update tag to mysql-driver in values.yaml

  3. Run a DolphinScheduler release in Kubernetes (See Installing the Chart)

  4. Add a MySQL datasource in Datasource manage

How to support Oracle datasource in Datasource manage?

Because of the commercial license, we cannot directly use the driver of Oracle.

If you want to add Oracle datasource, you can build a new image based on the apache/dolphinscheduler image as follows.

  1. Download the Oracle driver ojdbc8.jar (such as ojdbc8-19.9.0.0.jar)

  2. Create a new Dockerfile to add Oracle driver:

FROM apache/dolphinscheduler:latest
COPY ojdbc8-19.9.0.0.jar /opt/dolphinscheduler/lib
  1. Build a new docker image including Oracle driver:
docker build -t apache/dolphinscheduler:oracle-driver .
  1. Push the docker image apache/dolphinscheduler:oracle-driver to a docker registry

  2. Modify image repository and update tag to oracle-driver in values.yaml

  3. Run a DolphinScheduler release in Kubernetes (See Installing the Chart)

  4. Add a Oracle datasource in Datasource manage

For more information please refer to the incubator-dolphinscheduler documentation.