[ZEPPELIN-3840] Zeppelin on Kubernetes #3240

Leemoonsoo · 2018-11-24T10:33:38Z

What type of PR is it?

This PR adds ability to run Zeppelin on Kubernetes. It aims

Zero configuration to start Zeppelin on Kubernetes. (and Spark on Kubernetes)
Run everything on Kubernetes: Zeppelin, Interpreters, Spark.
Highly customizable to adopt various user configurations and extensions.

Key features are

Provides zeppelin-server.yaml file for kubectl to run Zeppelin server
All interpreters are automatically running as a Pod.
Spark interpreter automatically configured to use Spark on Kubernetes
Reverse proxy is configured to access Spark UI

To do

Document how reverse proxy for Spark UI works and how to configure custom domain.
Document how to customize zeppelin-server and interpreter yaml.
Document new configurations
Document how to mount volume for notebook and configurations

How it works

Run Zeppelin Server on Kubernetes

k8s/zeppelin-server.yaml is provided to run Zeppelin Server with few sidecars and configurations.
This file is easy to publish (user can easily consume it using curl), highly customizable while it includes all the necessary things.

K8s Interpreter launcher

This PR adds new module, launcher-k8s-standard under zeppelin/zeppelin-plugins/launcher/k8s-standard/ directory. This launcher is automatically being selected when Zeppelin is running on Kubernetes. The launcher both handles Spark interpreter and All other interpreters.

The launcher launches interpreter as a Pod using template k8s/interpreter/100-interpreter-pod.yaml.
Reason filename has 100- in prefix is because all files in the directory is consumed in alphabetical order by launcher on interpreter start/stop. User can drop more files here to extend/customize interpreter, and filename can be used to control order. The template is rendered by jinjava.

Spark interpreter

When interpreter group is spark, K8sRemoteInterpreterProcess sets necessary spark configuration automatically to use Spark on Kubernetes. User doesn't have to configure anything. It uses client mode.

Spark UI

We may make user manually configure port-forward or do something to access Spark UI, but that's not optimal. It is the best when Spark UI is automatically accessible when user have access to Zeppelin UI, without any extra configuration.

To enable this, Zeppelin server Pod has a reverse proxy as a sidecar, and it split traffic to Zeppelin server and Spark UI running in the other Pod. It assume both service.domain.com and *.service.domain.com point the nginx proxy address. service.domain.com is directed to ZeppelinServer, *.service.domain.com is directed to interpreter Pod.

<port>-<interpreter pod svc name>.service.domain.com is convention to access any application running in interpreter Pod. If Spark interpreter Pod is running with a name spark-axefeg and Spark UI is running on port 4040,

4040-spark-axefeg.service.domain.com

is the address to access Spark UI. Default service domain is local.zeppelin-project.org:8080, while local.zeppelin-project.org and *.local.zeppelin-project.org point 127.0.0.1, and it works with kubectl port-forward.

What is the Jira issue?

https://issues.apache.org/jira/browse/ZEPPELIN-3840

How should this be tested?

Prepare a Kubernetes cluster with enough resources (cpus > 5, mem > 6g).
If you're using minikube, check your capacity using kubectl describe node command before start.

You'll need to build Zeppelin docker image and Spark docker image to test. Please follow guide docs/quickstart/kubernetes.md.

To quickly try without building docker images, I have uploaded pre-built image on docker hub moon/zeppelin:0.9.0-SNAPSHOT, moon/spark:2.4.0. Try following command

ZEPPELIN_SERVER_YAML="curl -s https://raw.githubusercontent.com/Leemoonsoo/zeppelin/kubernetes/k8s/zeppelin-server.yaml"
$ZEPPELIN_SERVER_YAML | sed 's/apache\/zeppelin:0.9.0-SNAPSHOT/moon\/zeppelin:0.9.0-SNAPSHOT/' | sed 's/spark:2.4.0/moon\/spark:2.4.0/' | kubectl apply -f -

And port forward

kubectl port-forward zeppelin-server 8080:80

And browse http://localhost:8080

To clean up

$ZEPPELIN_SERVER_YAML | sed 's/apache\/zeppelin:0.9.0-SNAPSHOT/moon\/zeppelin:0.9.0-SNAPSHOT/' | sed 's/spark:2.4.0/moon\/spark:2.4.0/' | kubectl delete -f -

Screenshots (if appropriate)

See this video https://youtu.be/7E4ZGn4pnTo

Future work

Per interpreter docker image
Blocking communication between interpreter Pod.
Spark Interpreter Pod has Role CRUD for any pod/service in the same namespace. Which should be restricted to only Spark executors Pod.
Per note interpreter mode by default when Zeppelin is running on Kubernetes

Questions:

Does the licenses files need update? no
Is there breaking changes for older versions? no
Does this needs documentation? yes

...dard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java

...ns/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/Kubectl.java

Leemoonsoo · 2018-11-24T23:43:55Z

One question, there is no point using this if I have a Spark clster on YARN (Cloudera/Hortonworks) right? The way I see it this is focusing on having The Spark coming with Zeppelin now be able to lunch in a cluster by using KB. Am I right?

Right. I just pushed a commit fa36c18, and it allows user override master spark interpreter property. If master is not starting with k8s://, it does not configure Spark on Kubernetes. So I think this opens a possibility to connect other types of Spark cluster.

docs/quickstart/kubernetes.md

nrchakradhar · 2018-11-26T12:50:23Z

docs/quickstart/kubernetes.md

+$ docker build -t <tag> .
+```
+
+Finally, set custom image `<tag>` just created to `image` and `ZEPPELIN_K8S_CONTAINER_IMAGE` env variable of `zeppelin-server` container spec in `zeppelin-server.yaml` file.


Probably one small section that lists the current limitations will be useful. The one which is probable worth highlighting is about the container image used by the interpreter other than spark. It will be the same as Zeppelin container image.

Added more information below.

This information is also very helpful

nrchakradhar · 2018-11-26T12:52:33Z

k8s/interpreter/100-interpreter-pod.yaml

@@ -0,0 +1,143 @@
+#


The name of this file mentions "pod". But the yaml contains other resources like svc, rbac. Probably a better name can be used like 100-interpreter-spec.yaml?

File renamed.

nrchakradhar · 2018-11-26T13:03:54Z

k8s/zeppelin-server.yaml

+    - name: MASTER   # default value of master property for spark interpreter.
+      value: k8s://https://kubernetes.default.svc
+  - name: zeppelin-server-gateway
+    image: nginx:1.14.0


Is it better to create a svc for spark and use native Kubernetes redirection instead of ngnix as gateway?

What i'm trying to do here, make Spark-ui accessible when user have access to Zeppelin ui, without any extra configuration/command to run.

I'm not familiar with native Kubernetes redirection. Could you explain how it can be possible with native Kubernetes redirection?

Will this work for all modes (global, scoped, per-note, per-user)? I will try to check.
Implementing any security with embedded gateway might be a concern.

After some exploration, the change required will not be as simple as what has been provided. The approach could be as follows: (We have used it for other applications)

For the Spark-Driver UI, create a service with type=ClusterIP, port=randomPort, targetPort=driverUI port (4040 default).

Create an Ingress resource with path, service and port. Some annotations might be ingress controller specific sometimes. OR

With this the spark-driver UI is accessible directly at the K8S edge-node/ 'path'.

I guess this would require some customization to be done by the users anyway.
If I get a chance to explore other ways, I will get back.

This will work for all modes. In terms of security, while access to Spark UI will be through domain 4040-<Spark Interpreter pod name>.<service domain>, I would say it's safe while Pod name of Spark Interpreter is unpredictable.

Speaking of security, there're other aspect to consider.

Blocking communication between interpreter Pod.

Spark Interpreter Pod has Role CRUD for any pod/service in the same namespace. Which should be restricted to only Spark executors Pod. Otherwise SparkInterpreter Pod will able to access/terminate other Interpreter Pods.

I'd like to address above in separate PRs.

nrchakradhar · 2018-11-26T13:06:32Z

...dard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java

+  private final Map<String, String> envs;
+  private final String zeppelinServiceHost;
+  private final String zeppelinServiceRpcPort;
+  private final int podCreateTimeoutSec = 180;


Sometimes image pulling takes longer. Its better if its configurable.

Changed the code just use interpreter connection timeout for waiting pod creation.
Which is already configurable and designed to wait

interpreter process creation

interpreter process ready and open rpc server port

And i think it make sense to use this value to wait

interpreter pod creation

interpreter process ready and open rpc server port.

Thanks for this change. This makes users more comfortable :)

nrchakradhar · 2018-11-26T13:08:03Z

...dard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java

+    this.zeppelinServiceRpcPort = zeppelinServiceRpcPort;
+    this.portForward = portForward;
+    this.sparkImage = sparkImage;
+    this.podName = interpreterGroupName.toLowerCase() + "-" + getRandomString(6);


Is it better to use timestamp instead of RandomString?

Could you explain a bit more why timestamp can be better?

This was just a thought: string might collide sometimes while timestamp would be less prone for collision.

While code generating 6 char long random string using lower case alphabet, probability of collision is <number of running interpreter pods>/26^6 = <number of running interpreter pods>/308,915,776. I think probability of collision is small enough in practical.

One worry using timestamp in pod name is, pod name becomes predictable and that makes accessing to spark-ui insecure, while spark-ui is accessed through subdomain which is constructed based on interpreter pod name.

Agree. Let it be random string.

nrchakradhar · 2018-11-26T13:16:09Z

...dard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java

+    StringBuilder options = new StringBuilder();
+
+    options.append(" --master k8s://https://kubernetes.default.svc");
+    options.append(" --deploy-mode client");


With this Cluster mode for spark may not work. Though it may not be needed since the spark-interpreter itself is launched in a new pod. Usually driver memory might require changes as per application being run.

Updated to add --driver-memory configuration when spark.driver.memory is defined.

nrchakradhar · 2018-11-26T13:21:12Z

...d/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sStandardInterpreterLauncher.java

+   * Check if i'm running inside of kubernetes or not.
+   * @return
+   */
+  boolean isRunningOnKubernetes() {


Is this really needed? It appears redundant since the same check is happening in
InterpreterSetting.java

Yes, it is. I added a comment in the source code why we need isRunningOnKubernetes() inside of K8sStandardInterpreterLauncher, although K8sStandardInterpreterLauncher supposed to be used only in Kubernetes cluster which InterpreterSetting supposed to detect beforehand.

Thanks for the explanation. It is helpful.

nrchakradhar · 2018-11-26T13:24:47Z

...ns/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/Kubectl.java

@@ -0,0 +1,157 @@
+/*


Is it possible that we avoid kubectl binary dependency using io fabric java apis?
This could possibly be a separate PR. Please let me know if you want me to create a PR?

if kubernetes java client from fabric can handle kubernetes yaml spec file the same way kubectl handles, why not? I tried to use official kubernetes java client first, and changed to use kubectl because it does not have kubectl apply -f equivalent functions.

Usually with fabric8 API, applying of each resources has been tried by us. We will explore the usage of applying all resources. Initial reading of the code looks to be supporting it. However we will try and get back.

@Leemoonsoo
My colleague has put an example of the possible replacement under
zep-k8sctl

@nrchakradhar Cool! Thanks for sharing.

One question,

Let's say user is going to customize zeppelin-server.yaml, and interpreter.yaml. To mount volumns, add network restrictions, etc.

Therefore user might add additional sections in yaml, such as PersistentVolumeClaim, Secret, ConfigMap and so on. Also Kubernetes allows CRD, so there're some type of resource we don't know at this point.

But this code looks like need a modification if there's a new type of resource to apply. Will there any any solution for this?

@Leemoonsoo
Sorry for the delay in responding to this comment. Was busy in other activities.
We have tried multiple options to have functionality similar to kubectl apply, but fabric8 is having some bugs/issues/enhancements. Please refer the following Stackoverflow query for details:

For now we can have the current implementation and when fabric8 changes are available we can probably revisit this.

The Issue with fabric8 library related to kubectl apply or delete of resources using a single API is working with #1300 which is merged to master branch.
When it is available in the next release of fabric8io/kubernetes-client we can integrate the same. Using the above fix from fabric8 the change has been validated in the master branch of zep-k8sctl.
Can you please let us know if we create new JIRA issues so that we follow up or we will follow some different procedure.

Great news. Please feel free to create a JIRA issue and let me know if anything I can help.

maziyarpanahi · 2018-11-26T15:49:09Z

Right. I just pushed a commit fa36c18, and it allows user override master spark interpreter property. If master is not starting with k8s://, it does not configure Spark on Kubernetes. So I think this opens a possibility to connect other types of Spark cluster.

Awesome! that is a great news. Just out of curiosity, does the Zeppelin server in KB must be the one which has acess to the Spark_Home or the Spark interpreter which is now a pod? (Just to manage which one should be a gateway in Cloudera cluster).
Thanks again 👍

Leemoonsoo · 2018-11-27T03:09:49Z

Awesome! that is a great news. Just out of curiosity, does the Zeppelin server in KB must be the one which has acess to the Spark_Home or the Spark interpreter which is now a pod? (Just to manage which one should be a gateway in Cloudera cluster).

Spark Interpreter is running inside of its own Pod. And Spark driver is running in Spark Interpreter. So Spark Interpreter Pod is the one, who need SPARK_HOME and access from/to Spark master/executors in your cluster.

Tagar · 2018-11-28T18:44:41Z

Would be great to cut 0.9 release with this and all other great improvements that are currently only available in master. So more folks can take leverage of those features, and provide feedback sooner.

Thank you!

xunliu · 2018-11-29T01:52:20Z

zeppelin-interpreter/src/main/java/org/apache/zeppelin/conf/ZeppelinConfiguration.java

@@ -816,6 +840,13 @@ public int getClusterHeartbeatTimeout() {
    ZEPPELIN_CLUSTER_HEARTBEAT_INTERVAL("zeppelin.cluster.heartbeat.interval", 3000),
    ZEPPELIN_CLUSTER_HEARTBEAT_TIMEOUT("zeppelin.cluster.heartbeat.timeout", 9000),

+    ZEPPELIN_K8S_MODE("zeppelin.k8s.mode", "auto"),              // auto | on | off


Later, zeppelin may also support YARN. Now YARN also supports Docker. Is it better to change this configuration to ZEPPELIN_RUN_MODE?
For example: ZEPPELIN_RUN_MODE("zeppelin.run.mode", "auto"), // auto | k8s | yarn | normal

Make sense. Updated.

whai2kay · 2018-12-10T10:15:40Z

Right. I just pushed a commit fa36c18, and it allows user override master spark interpreter property. If master is not starting with k8s://, it does not configure Spark on Kubernetes. So I think this opens a possibility to connect other types of Spark cluster.

It would seem that when setting the master to: spark://spark-cluster:7077 , it would just use local[*] instead. Is connecting to a cluster using client mode not yet supported?

EDIT: Adding an additional field to the spark interpreter conf "spark.master" and setting that to the spark cluster master url works (eg: spark://spark-cluster:7077)

Leemoonsoo · 2019-01-15T23:13:33Z

Sorry for responding late. Thanks @whai2kay for testing it out and confirming configure spark using "spark.master" interpreter conf.

This is just an initial implementation and it can be a starting point to improvements from applying Kubernetes client library to network policy for security, and so on.

I'd like to merge this PR to master, so we can start working on further improvements, If there's no more comments.

ajaygk95 · 2019-01-17T07:57:19Z

...ns/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/Kubectl.java

+  }
+
+  public String apply(String spec) throws IOException {
+    return execAndGet(new String[]{"apply", "-f", "-"}, spec);


kubectl apply -f does not work if there is a generateName in the interpreter spec yaml. Its better to use kubectl create -f if resources are being created. Please find the discussion here.

I didn't know about generateName. Thanks for pointing this out with discussion link.
I agree using create -f. While this PR works itself, I'd like to merge this first and then follow up apply -> create in a separate PR. Does it sounds like a plan?

Created https://issues.apache.org/jira/browse/ZEPPELIN-3954. I'll follow up this issue.

chrissng · 2019-05-03T02:25:21Z

...dard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java

+      options.append(" --driver-memory " + properties.get("spark.driver.memory"));
+    }
+    options.append(" --conf spark.kubernetes.namespace=" + kubectl.getNamespace());
+    options.append(" --conf spark.executor.instances=1");


@Leemoonsoo Are there alternative means to specify spark.executor.instances? I don't see a means to do so as this function build the spark submit options and is appended to the end. i.e. these will override all other options as specified in https://github.com/apache/zeppelin/pull/3240/files#diff-6d1d3084f55bdd519e39ede4a619e73dR269

%spark.conf spark.executor.instances <number>

inside a note will allow configuring this value.
But right, it'll be better not override SPARK_SUBMIT_OPTIONS.

The default value of spark.executor.instances is 2. And I think it's okay to remove this line and use default value.

chrissng · 2019-05-03T02:30:16Z

...d/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sStandardInterpreterLauncher.java

+  private String getZeppelinServiceHost() throws IOException {
+    if (isRunningOnKubernetes()) {
+      return String.format("%s.%s.svc.cluster.local",
+              getHostname(), // service name and pod name should be the same


@Leemoonsoo Great work on making Zeppelin work in Kubernetes!

I have a concern with this approach of matching Service and Pod names via hostname. This will not work for zeppelin pods that are provisioned via Deployments or StatefulSets as pod names are dynamically generated during creation.

Good point. Notebook serving (#3356) has an exact problem while TestTask runs Zeppelin using Job, and ServingTask runs Zeppelin using Deployment.

In serving implementation, K8sStandardInterpreterLauncher has modified to looking for SERVICE_NAME env variable. Check https://github.com/apache/zeppelin/pull/3356/files#diff-dfad7c26921ec97aa7bca7fc3180408fL109. Does it make sense?

That should work 👍

chrissng · 2019-05-03T03:45:59Z

...ns/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/Kubectl.java

+    CommandLine cmd = new CommandLine(kubectlCmd);
+    cmd.addArguments(args);
+
+    ExecuteWatchdog watchdog = new ExecuteWatchdog(60 * 1000);


@Leemoonsoo This value should be the same as the kubectl timeout flags. In the case of kubectl wait: https://github.com/apache/zeppelin/pull/3240/files#diff-7b231819fa2236eab55c934add3b56e8R69

This timeout may be reached before kubectl wait's timeout, causing an exception to be thrown:

INFO [2019-05-03 03:24:59,232] ({SchedulerFactory3} Kubectl.java[execAndGet]:121) - kubectl [wait, pod/spark-foqspk, --for=condition=Ready, --timeout=600s, --namespace=aikwei-sng] WARN [2019-05-03 03:25:59,257] ({SchedulerFactory3} NotebookServer.java[onStatusChange]:1671) - Job paragraph_1556788652522_1837703941 is finished, status: ERROR, exception: null, result: %text org.apache.zeppelin.interpreter.InterpreterException: java.io.IOException: at org.apache.zeppelin.interpreter.remote.RemoteInterpreter.open(RemoteInterpreter.java:134) at org.apache.zeppelin.interpreter.remote.RemoteInterpreter.getFormType(RemoteInterpreter.java:298) at org.apache.zeppelin.notebook.Paragraph.jobRun(Paragraph.java:417) at org.apache.zeppelin.notebook.Paragraph.jobRun(Paragraph.java:76) at org.apache.zeppelin.scheduler.Job.run(Job.java:172) at org.apache.zeppelin.scheduler.AbstractScheduler.runJob(AbstractScheduler.java:121) at org.apache.zeppelin.scheduler.RemoteScheduler$JobRunner.run(RemoteScheduler.java:187) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) at java.util.concurrent.FutureTask.run(FutureTask.java:266) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180) at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748) Caused by: java.io.IOException: at org.apache.zeppelin.interpreter.launcher.Kubectl.execAndGet(Kubectl.java:141) at org.apache.zeppelin.interpreter.launcher.Kubectl.execAndGet(Kubectl.java:106) at org.apache.zeppelin.interpreter.launcher.Kubectl.wait(Kubectl.java:65) at org.apache.zeppelin.interpreter.launcher.K8sRemoteInterpreterProcess.start(K8sRemoteInterpreterProcess.java:92) at org.apache.zeppelin.interpreter.ManagedInterpreterGroup.getOrCreateInterpreterProcess(ManagedInterpreterGroup.java:65) at org.apache.zeppelin.interpreter.remote.RemoteInterpreter.getOrCreateInterpreterProcess(RemoteInterpreter.java:110) at org.apache.zeppelin.interpreter.remote.RemoteInterpreter.internal_create(RemoteInterpreter.java:163) at org.apache.zeppelin.interpreter.remote.RemoteInterpreter.open(RemoteInterpreter.java:131) ... 13 more Caused by: org.apache.commons.exec.ExecuteException: Process exited with an error: 143 (Exit value: 143) at org.apache.commons.exec.DefaultExecutor.executeInternal(DefaultExecutor.java:404) at org.apache.commons.exec.DefaultExecutor.execute(DefaultExecutor.java:166) at org.apache.commons.exec.DefaultExecutor.execute(DefaultExecutor.java:153) at org.apache.zeppelin.interpreter.launcher.Kubectl.execute(Kubectl.java:155) at org.apache.zeppelin.interpreter.launcher.Kubectl.execAndGet(Kubectl.java:125) ... 20 more

Thanks for pointing it out!

Leemoonsoo and others added 27 commits November 22, 2018 13:05

add k8s-standard launcher module

d2f3d5b

kubectl with exec

07489f7

K8sRemoteInterpreterProcess

5f602a6

add k8s dir in package

52bb6c7

correct plugin name

36cf391

add rbac

58f9f19

kubernetes mode configuration

2fd2ac8

run kubernetes launcher

9f1b7a1

create and connect interpreter pod

0dea383

print spec file contents on debug log

18b8f68

add services on RBAC

86e8764

interpreter pod cascade delete on zeppelin-server delete

7fe9823

use headless service for interpreter pod

263d859

configure spark on kubernetes

7a87367

add signal handler

a4072e6

configure namespace

2960dcb

Rbac role, rolebinding

b0e2c36

load properties and environment variables

0d472ea

make spark container image configurable

f4166ad

let user override namespace

2b579ff

rename file

f305611

add license

0f7c0d4

install kubectl and configure log4j in docker image

9341fcb

spark ui support

3078bac

add test

ec09b8b

update dockerfile

e9ce64f

Add docs

64a56b5

Leemoonsoo force-pushed the kubernetes branch from f02151d to 64a56b5 Compare November 24, 2018 10:39

Leemoonsoo closed this Nov 24, 2018

Leemoonsoo reopened this Nov 24, 2018

xunliu reviewed Nov 24, 2018

View reviewed changes

...dard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java Outdated Show resolved Hide resolved

xunliu reviewed Nov 24, 2018

View reviewed changes

...ns/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/Kubectl.java Outdated Show resolved Hide resolved

update corresponding test as well

11960dd

logger -> LOGGER

22e090f

nrchakradhar reviewed Nov 26, 2018

View reviewed changes

docs/quickstart/kubernetes.md Outdated Show resolved Hide resolved

nrchakradhar reviewed Nov 26, 2018

View reviewed changes

Leemoonsoo added 2 commits November 26, 2018 18:42

address code review

9ab6fc4

document configurations

993a0e4

localtest.me -> local.zeppelin-project.org

4e7d817

xunliu reviewed Nov 29, 2018

View reviewed changes

Leemoonsoo and others added 2 commits November 30, 2018 11:41

zeppelin.k8s.mode -> zeppelin.run.mode

423412a

update how it works on docs, add some comments on yaml files

0100a36

ajaygk95 reviewed Jan 17, 2019

View reviewed changes

asfgit closed this in b13651c Jan 18, 2019

chrissng reviewed May 3, 2019

View reviewed changes

[ZEPPELIN-3840] Zeppelin on Kubernetes #3240

[ZEPPELIN-3840] Zeppelin on Kubernetes #3240

Conversation

Leemoonsoo commented Nov 24, 2018 • edited Loading

What type of PR is it?

How it works

Run Zeppelin Server on Kubernetes

K8s Interpreter launcher

Spark interpreter

Spark UI

What is the Jira issue?

How should this be tested?

Screenshots (if appropriate)

Future work

Questions:

Leemoonsoo commented Nov 24, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Leemoonsoo Nov 28, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Leemoonsoo Dec 5, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

maziyarpanahi commented Nov 26, 2018

Leemoonsoo commented Nov 27, 2018

Tagar commented Nov 28, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

whai2kay commented Dec 10, 2018 • edited Loading

Leemoonsoo commented Jan 15, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ajaygk95 Jan 18, 2019 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Leemoonsoo commented Nov 24, 2018 •

edited

Loading

Leemoonsoo Nov 28, 2018 •

edited

Loading

Leemoonsoo Dec 5, 2018 •

edited

Loading

whai2kay commented Dec 10, 2018 •

edited

Loading

ajaygk95 Jan 18, 2019 •

edited

Loading