suneet-s commented on code in PR #14674:
URL: https://github.com/apache/druid/pull/14674#discussion_r1278114114
##########
website/static/css/custom.css:
##########
@@ -99,15 +99,3 @@ article iframe {
margin-right: auto;
max-width: 100%;
}
-.getAPI {
- color: #0073e6;
- font-weight: bold;
-}
-.postAPI {
- color: #00bf7d;
- font-weight: bold;
-}
-.deleteAPI {
- color: #f49200;
- font-weight: bold;
-}
Review Comment:
accidental change?
##########
extensions-contrib/kubernetes-overlord-extensions/README.md:
##########
@@ -0,0 +1,32 @@
+<!--
+ ~ Licensed to the Apache Software Foundation (ASF) under one
+ ~ or more contributor license agreements. See the NOTICE file
+ ~ distributed with this work for additional information
+ ~ regarding copyright ownership. The ASF licenses this file
+ ~ to you under the Apache License, Version 2.0 (the
+ ~ "License"); you may not use this file except in compliance
+ ~ with the License. You may obtain a copy of the License at
+ ~
+ ~ http://www.apache.org/licenses/LICENSE-2.0
+ ~
+ ~ Unless required by applicable law or agreed to in writing,
+ ~ software distributed under the License is distributed on an
+ ~ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ ~ KIND, either express or implied. See the License for the
+ ~ specific language governing permissions and limitations
+ ~ under the License.
+ -->
+
+druid-kubernetes-overlord-extensions
+=============
+
+Overview
+=============
+The Kubernetes Task Scheduling extension allows a Druid cluster running on
Kubernetes to schedule
+its tasks as Kubernetes Jobs instead of sending them to workers (middle
managers or indexers).
+
+Documentation
+=============
+See the druid.apache.org website or under [Druid GitHub
Repo](https://github.com/apache/druid/blob/master/docs/development/extensions-contrib/k8s-jobs.md).
Review Comment:
```suggestion
More detailed documentation about how to configure and use the extension is
available [here](../../docs/development/extensions-contrib/k8s-jobs.md)
```
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
Review Comment:
```suggestion
Task specific pod templates can be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
```
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
+ labels:
+ custom-label: "hello"
+ spec:
+ affinity: {}
+ containers:
+ - command:
+ - sh
+ - -c
+ - |
+ /peon.sh /druid/data 1
+ env:
+ - name: CUSTOM_ENV_VARIABLE
+ value: "hello"
+ image: apache/druid:26.0.0
+ name: main
+ ports:
+ - containerPort: 8091
+ name: druid-tls-port
+ protocol: TCP
+ - containerPort: 8100
+ name: druid-port
+ protocol: TCP
+ resources:
+ limits:
+ cpu: "1"
+ memory: 2400M
+ requests:
+ cpu: "1"
+ memory: 2400M
+ volumeMounts:
+ - mountPath: /opt/druid/conf/druid/cluster/master/coordinator-overlord #
runtime props are still mounted in this location because that's where peon.sh
looks for configs
+ name: nodetype-config-volume
+ readOnly: true
+ - mountPath: /druid/data
+ name: data-volume
+ - mountPath: /druid/deepstorage
+ name: deepstorage-volume
+ restartPolicy: "Never"
+ securityContext:
+ fsGroup: 1000
+ runAsGroup: 1000
+ runAsUser: 1000
+ tolerations:
+ - effect: NoExecute
+ key: node.kubernetes.io/not-ready
+ operator: Exists
+ tolerationSeconds: 300
+ - effect: NoExecute
+ key: node.kubernetes.io/unreachable
+ operator: Exists
+ tolerationSeconds: 300
+ volumes:
+ - configMap:
+ defaultMode: 420
+ name: druid-tiny-cluster-peons-config
+ name: nodetype-config-volume
+ - emptyDir: {}
+ name: data-volume
+ - emptyDir: {}
+ name: deepstorage-volume
+```
+
+The below runtime properties need to be passed to the Job's peon process.
```
-apiVersion: rbac.authorization.k8s.io/v1
+druid.port=8100 (what port the peon should run on)
+druid.peon.mode=remote
+druid.service=druid/peon (for metrics reporting)
+druid.indexer.task.baseTaskDir=/druid/data (this should match the argument to
the ./peon.sh run command in the PodTemplate)
+druid.indexer.runner.type=k8s
+druid.indexer.task.encapsulatedTask=true
+```
+
+Any runtime property or JVM config used by the peon process can also be
passed. E.G. below is a example of a ConfigMap that can be used to generate the
`nodetype-config-volume` mount in the above template.
+```
+kind: ConfigMap
+metadata:
+ name: druid-tiny-cluster-peons-config
+ namespace: default
+apiVersion: v1
+data:
+ jvm.config: |-
+ -server
+ -XX:MaxDirectMemorySize=1000M
+ -Duser.timezone=UTC
+ -Dfile.encoding=UTF-8
+ -Dlog4j.debug
+ -Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
+ -Djava.io.tmpdir=/druid/data
+ -Xmx1024M
+ -Xms1024M
+ log4j2.xml: |-
+ <?xml version="1.0" encoding="UTF-8" ?>
+ <Configuration status="WARN">
+ <Appenders>
+ <Console name="Console" target="SYSTEM_OUT">
+ <PatternLayout pattern="%d{ISO8601} %p [%t] %c - %m%n"/>
+ </Console>
+ </Appenders>
+ <Loggers>
+ <Root level="info">
+ <AppenderRef ref="Console"/>
+ </Root>
+ </Loggers>
+ </Configuration>
+ runtime.properties: |
+ druid.port=8100
+ druid.service=druid/peon
+ druid.server.http.numThreads=5
+ druid.indexer.task.baseTaskDir=/druid/data
+ druid.indexer.runner.type=k8s
+ druid.peon.mode=remote
+ druid.indexer.task.encapsulatedTask=true
+```
+
+
+Additional Configuration
+### Properties
+|Property| Possible Values | Description
|Default|required|
+|--------|-----------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------|--------|
+|`druid.indexer.runner.debugJobs`| `boolean` | Clean up K8s jobs after
tasks complete.
|False|No|
+|`druid.indexer.runner.sidecarSupport`| `boolean` | Deprecated, specify
adapter type as runtime property `druid.indexer.runner.k8s.adapter.type:
overlordMultiContainer` instead. If your overlord pod has sidecars, this will
attempt to start the task with the same sidecars as the overlord pod. |False|No|
+|`druid.indexer.runner.primaryContainerName`| `String` | If running
with sidecars, the `primaryContainerName` should be that of your druid
container like `druid-overlord`.
|First container in `podSpec` list|No|
+|`druid.indexer.runner.kubexitImage`| `String` | Used kubexit project
to help shutdown sidecars when the main pod completes. Otherwise jobs with
sidecars never terminate.
|karlkfi/kubexit:latest|No|
+|`druid.indexer.runner.disableClientProxy`| `boolean` | Use this if you
have a global http(s) proxy and you wish to bypass it.
|false|No|
+|`druid.indexer.runner.maxTaskDuration`| `Duration` | Max time a task is
allowed to run for before getting killed
|`PT4H`|No|
+|`druid.indexer.runner.taskCleanupDelay`| `Duration` | How long do jobs
stay around before getting reaped from K8s
|`P2D`|No|
+|`druid.indexer.runner.taskCleanupInterval`| `Duration` | How often to
check for jobs to be reaped
|`PT10M`|No|
+|`druid.indexer.runner.K8sjobLaunchTimeout`| `Duration` | How long to
wait to launch a K8s task before marking it as failed, on a resource
constrained cluster it may take some time.
|`PT1H`|No|
+|`druid.indexer.runner.javaOptsArray`| `JsonArray` | java opts for the
task.
|`-Xmx1g`|No|
+|`druid.indexer.runner.labels`| `JsonObject` | Additional labels you want
to add to peon pod
|`{}`|No|
+|`druid.indexer.runner.annotations`| `JsonObject` | Additional annotations
you want to add to peon pod
|`{}`|No|
+|`druid.indexer.runner.peonMonitors`| `JsonArray` | Overrides
`druid.monitoring.monitors`. Use this property if you don't want to inherit
monitors from the Overlord.
|`[]`|No|
+|`druid.indexer.runner.graceTerminationPeriodSeconds`| `Long` |
Number of seconds you want to wait after a sigterm for container lifecycle
hooks to complete. Keep at a smaller value if you want tasks to hold locks for
shorter periods.
|`PT30S` (K8s default)|No|
+|`druid.indexer.runner.capacity`| `Integer` | Number of concurrent jobs
that can be sent to Kubernetes.
|`2147483647 `|No|
Review Comment:
```suggestion
|`druid.indexer.runner.capacity`| `Integer` | Number of concurrent
jobs that can be sent to Kubernetes.
|`2147483647`|No|
```
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
+ labels:
+ custom-label: "hello"
+ spec:
+ affinity: {}
+ containers:
+ - command:
+ - sh
+ - -c
+ - |
+ /peon.sh /druid/data 1
+ env:
+ - name: CUSTOM_ENV_VARIABLE
+ value: "hello"
+ image: apache/druid:26.0.0
+ name: main
+ ports:
+ - containerPort: 8091
+ name: druid-tls-port
+ protocol: TCP
+ - containerPort: 8100
+ name: druid-port
+ protocol: TCP
+ resources:
+ limits:
+ cpu: "1"
+ memory: 2400M
+ requests:
+ cpu: "1"
+ memory: 2400M
+ volumeMounts:
+ - mountPath: /opt/druid/conf/druid/cluster/master/coordinator-overlord #
runtime props are still mounted in this location because that's where peon.sh
looks for configs
+ name: nodetype-config-volume
+ readOnly: true
+ - mountPath: /druid/data
+ name: data-volume
+ - mountPath: /druid/deepstorage
+ name: deepstorage-volume
+ restartPolicy: "Never"
+ securityContext:
+ fsGroup: 1000
+ runAsGroup: 1000
+ runAsUser: 1000
Review Comment:
Can you explain what these settings are please
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
Review Comment:
Can you provide a real world example of an annotation that would be helpful
in the template. Same comment about the labels.
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -28,85 +28,45 @@ Consider this an [EXPERIMENTAL](../experimental.md) feature
mostly because it ha
## How it works
-The K8s extension builds a pod spec using the specified pod adapter, the
default implementation takes the podSpec of your `Overlord` pod and creates a
kubernetes job from this podSpec. Thus if you have sidecars such as Splunk or
Istio it can optionally launch a task as a K8s job. All jobs are natively
restorable, they are decoupled from the druid deployment, thus restarting pods
or doing upgrades has no affect on tasks in flight. They will continue to run
and when the overlord comes back up it will start tracking them again.
+The K8s extension builds a pod spec from each task using the specified pod
adapter. All jobs are natively restorable, they are decoupled from the druid
deployment, thus restarting pods or doing upgrades has no affect on tasks in
flight. They will continue to run and when the overlord comes back up it will
start tracking them again.
Review Comment:
```suggestion
The K8s extension builds a pod spec for each task using the specified pod
adapter. All jobs are natively restorable, they are decoupled from the Druid
deployment, thus restarting pods or doing upgrades has no affect on tasks in
flight. They will continue to run and when the overlord comes back up it will
start tracking them again.
```
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
+ labels:
+ custom-label: "hello"
+ spec:
+ affinity: {}
+ containers:
+ - command:
+ - sh
+ - -c
+ - |
+ /peon.sh /druid/data 1
+ env:
+ - name: CUSTOM_ENV_VARIABLE
+ value: "hello"
+ image: apache/druid:26.0.0
+ name: main
+ ports:
+ - containerPort: 8091
+ name: druid-tls-port
+ protocol: TCP
+ - containerPort: 8100
+ name: druid-port
+ protocol: TCP
+ resources:
+ limits:
+ cpu: "1"
+ memory: 2400M
+ requests:
+ cpu: "1"
+ memory: 2400M
+ volumeMounts:
+ - mountPath: /opt/druid/conf/druid/cluster/master/coordinator-overlord #
runtime props are still mounted in this location because that's where peon.sh
looks for configs
Review Comment:
This seems a little confusing. Should `peon.sh` to look for runtime
properties in another location for the peons?
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
+ labels:
+ custom-label: "hello"
+ spec:
+ affinity: {}
+ containers:
+ - command:
+ - sh
+ - -c
+ - |
+ /peon.sh /druid/data 1
+ env:
+ - name: CUSTOM_ENV_VARIABLE
+ value: "hello"
+ image: apache/druid:26.0.0
+ name: main
+ ports:
+ - containerPort: 8091
+ name: druid-tls-port
+ protocol: TCP
+ - containerPort: 8100
+ name: druid-port
+ protocol: TCP
+ resources:
+ limits:
+ cpu: "1"
+ memory: 2400M
+ requests:
+ cpu: "1"
+ memory: 2400M
Review Comment:
This requests seems high given that Xmx is 1G below. Does it need to be this
high?
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
+ labels:
+ custom-label: "hello"
+ spec:
+ affinity: {}
+ containers:
+ - command:
+ - sh
+ - -c
+ - |
+ /peon.sh /druid/data 1
+ env:
+ - name: CUSTOM_ENV_VARIABLE
+ value: "hello"
+ image: apache/druid:26.0.0
+ name: main
+ ports:
+ - containerPort: 8091
+ name: druid-tls-port
+ protocol: TCP
+ - containerPort: 8100
+ name: druid-port
+ protocol: TCP
+ resources:
+ limits:
+ cpu: "1"
Review Comment:
Do we need to set a limit?
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -115,33 +75,197 @@ You can keep your Dockerfile the same but you must have a
sidecar spec like so:
- arg2
```
-The following roles must also be accessible. An example spec could be:
+For both of these adapters, you can add optional labels to your K8s jobs /
pods if you need them by using the following configuration:
+`druid.indexer.runner.labels: '{"key":"value"}'`
+Annotations are the same with:
+`druid.indexer.runner.annotations: '{"key":"value"}'`
+
+All other configurations you had for the middle manager tasks must be moved
under the overlord with one caveat, you must specify javaOpts as an array:
+`druid.indexer.runner.javaOptsArray`, `druid.indexer.runner.javaOpts` is no
longer supported.
+
+If you are running without a middle manager you need to also use
`druid.processing.intermediaryData.storage.type=deepstore`
+
+### Custom Template Pod Adapter
+The custom template pod adapter allows you to specify a pod template file per
task type for more flexibility on how to define your pods. This adapter expects
a [Pod
Template](https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates) to
be available on the overlord's file system. This pod template is used as the
base of the pod spec for the Kubernetes Job. You can override things like
labels, environment variables, resources, annotation, or even the base image
with this template. To enable this pod adapter you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
+
+The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
+
+Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
+
+The following is an example Pod Template that uses the regular druid docker
image.
+```
+apiVersion: "v1"
+kind: "PodTemplate"
+template:
+ metadata:
+ annotations:
+ custom-annotation: "hello"
+ labels:
+ custom-label: "hello"
+ spec:
+ affinity: {}
+ containers:
+ - command:
+ - sh
+ - -c
+ - |
+ /peon.sh /druid/data 1
+ env:
+ - name: CUSTOM_ENV_VARIABLE
+ value: "hello"
+ image: apache/druid:26.0.0
Review Comment:
```suggestion
image: apache/druid:{{DRUIDVERSION}}
```
There is a script that runs and replaces {{DRUIDVERSION}} with the version
of the docs where this is published
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -28,85 +28,45 @@ Consider this an [EXPERIMENTAL](../experimental.md) feature
mostly because it ha
## How it works
-The K8s extension builds a pod spec using the specified pod adapter, the
default implementation takes the podSpec of your `Overlord` pod and creates a
kubernetes job from this podSpec. Thus if you have sidecars such as Splunk or
Istio it can optionally launch a task as a K8s job. All jobs are natively
restorable, they are decoupled from the druid deployment, thus restarting pods
or doing upgrades has no affect on tasks in flight. They will continue to run
and when the overlord comes back up it will start tracking them again.
+The K8s extension builds a pod spec from each task using the specified pod
adapter. All jobs are natively restorable, they are decoupled from the druid
deployment, thus restarting pods or doing upgrades has no affect on tasks in
flight. They will continue to run and when the overlord comes back up it will
start tracking them again.
-## Pod Adapters
-The logic defining how the pod template is built for your kubernetes job
depends on which pod adapter you have specified.
-
-### Overlord Single Container Pod Adapter
-The overlord single container pod adapter takes the podSpec of your `Overlord`
pod and creates a kubernetes job from this podSpec. This is the default pod
adapter implementation, to explicitly enable it you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: overlordSingleContainer`
-
-### Overlord Multi Container Pod Adapter
-The overlord multi container pod adapter takes the podSpec of your `Overlord`
pod and creates a kubernetes job from this podSpec. It uses kubexit to manage
dependency ordering between the main container that runs your druid peon and
other sidecars defined in the `Overlord` pod spec. To enable this pod adapter
you can specify the runtime property `druid.indexer.runner.k8s.adapter.type:
overlordMultiContainer`
-
-### Custom Template Pod Adapter
-The custom template pod adapter allows you to specify a pod template file per
task type. This adapter requires you to specify a `base` pod spec which will
be used in the case that a task specific pod spec has not been defined. To
enable this pod adapter you can specify the runtime property
`druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
-
-The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
-Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
## Configuration
To use this extension please make sure to
[include](../../configuration/extensions.md#loading-extensions)`druid-kubernetes-overlord-extensions`
in the extensions load list for your overlord process.
-The extension uses the task queue to limit how many concurrent tasks (K8s
jobs) are in flight so it is required you have a reasonable value for
`druid.indexer.queue.maxSize`. Additionally set the variable
`druid.indexer.runner.namespace` to the namespace in which you are running
druid.
+The extension uses `druid.indexer.runner.capacity` to limit the number of k8s
jobs in flight. You should set this to a reasonable value. Additionally set the
variable `druid.indexer.runner.namespace` to the namespace in which you are
running druid.
Review Comment:
```suggestion
The extension uses `druid.indexer.runner.capacity` to limit the number of
concurrent peons running in the Kubernetes cluster. Setting this value too high
can cause the overlord to be OOMKilled as it spawns more threads to keep track
of the peons.
```
> Additionally set the variable `druid.indexer.runner.namespace` to the
namespace in which you are running druid.
It feels like this belongs on the next line with the rest of the required
configuration
##########
docs/development/extensions-contrib/k8s-jobs.md:
##########
@@ -28,85 +28,45 @@ Consider this an [EXPERIMENTAL](../experimental.md) feature
mostly because it ha
## How it works
-The K8s extension builds a pod spec using the specified pod adapter, the
default implementation takes the podSpec of your `Overlord` pod and creates a
kubernetes job from this podSpec. Thus if you have sidecars such as Splunk or
Istio it can optionally launch a task as a K8s job. All jobs are natively
restorable, they are decoupled from the druid deployment, thus restarting pods
or doing upgrades has no affect on tasks in flight. They will continue to run
and when the overlord comes back up it will start tracking them again.
+The K8s extension builds a pod spec from each task using the specified pod
adapter. All jobs are natively restorable, they are decoupled from the druid
deployment, thus restarting pods or doing upgrades has no affect on tasks in
flight. They will continue to run and when the overlord comes back up it will
start tracking them again.
-## Pod Adapters
-The logic defining how the pod template is built for your kubernetes job
depends on which pod adapter you have specified.
-
-### Overlord Single Container Pod Adapter
-The overlord single container pod adapter takes the podSpec of your `Overlord`
pod and creates a kubernetes job from this podSpec. This is the default pod
adapter implementation, to explicitly enable it you can specify the runtime
property `druid.indexer.runner.k8s.adapter.type: overlordSingleContainer`
-
-### Overlord Multi Container Pod Adapter
-The overlord multi container pod adapter takes the podSpec of your `Overlord`
pod and creates a kubernetes job from this podSpec. It uses kubexit to manage
dependency ordering between the main container that runs your druid peon and
other sidecars defined in the `Overlord` pod spec. To enable this pod adapter
you can specify the runtime property `druid.indexer.runner.k8s.adapter.type:
overlordMultiContainer`
-
-### Custom Template Pod Adapter
-The custom template pod adapter allows you to specify a pod template file per
task type. This adapter requires you to specify a `base` pod spec which will
be used in the case that a task specific pod spec has not been defined. To
enable this pod adapter you can specify the runtime property
`druid.indexer.runner.k8s.adapter.type: customTemplateAdapter`
-
-The base pod template must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.base: /path/to/basePodSpec.yaml`
-Task specific pod templates must be specified as the runtime property
`druid.indexer.runner.k8s.podTemplate.{taskType}:
/path/to/taskSpecificPodSpec.yaml` where {taskType} is the name of the task
type i.e `index_parallel`
## Configuration
To use this extension please make sure to
[include](../../configuration/extensions.md#loading-extensions)`druid-kubernetes-overlord-extensions`
in the extensions load list for your overlord process.
-The extension uses the task queue to limit how many concurrent tasks (K8s
jobs) are in flight so it is required you have a reasonable value for
`druid.indexer.queue.maxSize`. Additionally set the variable
`druid.indexer.runner.namespace` to the namespace in which you are running
druid.
+The extension uses `druid.indexer.runner.capacity` to limit the number of k8s
jobs in flight. You should set this to a reasonable value. Additionally set the
variable `druid.indexer.runner.namespace` to the namespace in which you are
running druid.
Review Comment:
> You should set this to a reasonable value.
Can we include some guidance on what is reasonable. I think mentioning that
having this set to infinite means that the overlord will have more threads will
be useful for an operator to tune their cluster.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
