[
https://issues.apache.org/jira/browse/KYLIN-4447?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17092743#comment-17092743
]
Xiaoxiang Yu edited comment on KYLIN-4447 at 5/4/20, 7:57 AM:
--------------------------------------------------------------
h2. Explanation
# CuratorScheduler is used as default JobScheduler because it is more flexible.
# Spark building require use +cluster+ as +deployMode+. If you forget it, your
spark application will never submitted successfully because Hadoop cluster can
not resolve hostname of Pod (Spark Driver).
# To modify +/etc/hosts+ in Pod, please check this :
[https://kubernetes.io/docs/concepts/services-networking/add-entries-to-pod-etc-hosts-with-host-aliases/]
.
# To build you own *kylin-client* docker image, please don't forget to
download and put following jars into +KYLIN_HOME/tomcat/lib+ to enable *tomcat
session sharing*.
##
[https://repo1.maven.org/maven2/de/javakaffee/msm/memcached-session-manager-tc7/2.1.1/]
##
[https://repo1.maven.org/maven2/de/javakaffee/msm/memcached-session-manager/2.1.1/]
# If you have difficulty in configure *filebeat*, please check this
[https://www.elastic.co/guide/en/beats/filebeat/current/index.html] .
# External query cache is enabled by default, if you are interested in detail,
you may check
[http://kylin.apache.org/blog/2019/07/30/detailed-analysis-of-refine-query-cache/]
.
# Job Server and Query Server is separated into two service because job server
will reject SQL query request.
# All configuration files is separated from Docker image, please use configMap
or secert. Compared to confogMap, secrets is more recommended for security
reason.
# Some kylin-client image will be published to Dockerhub, here is the link
[https://hub.docker.com/r/apachekylin/kylin-client] . You may consider
contributed your Dokcerfile to kylin's repo if you are interested.
was (Author: hit_lacus):
h2. Explanation
# CuratorScheduler is used as default JobScheduler because it is more flexible.
# Spark building require use +cluster+ as +deployMode+. If you forget it, your
spark application will never submitted successfully because Hadoop cluster can
not resolve hostname of Pod (Spark Driver).
# To modify +/etc/hosts+ in Pod, please check this :
[https://kubernetes.io/docs/concepts/services-networking/add-entries-to-pod-etc-hosts-with-host-aliases/]
.
# To build you own *kylin-client* docker image, please don't forget to
download and put following jars into +KYLIN_HOME/tomcat/lib+ to enable *tomcat
session sharing*.
##
[https://repo1.maven.org/maven2/de/javakaffee/msm/memcached-session-manager-tc7/2.1.1/]
##
[https://repo1.maven.org/maven2/de/javakaffee/msm/memcached-session-manager/2.1.1/]
# If you have difficulty in configure *filebeat*, please check this
[https://www.elastic.co/guide/en/beats/filebeat/current/index.html] .
# External query cache is enabled by default, if you are interested in detail,
you may check
[http://kylin.apache.org/blog/2019/07/30/detailed-analysis-of-refine-query-cache/]
.
> Kylin on kubernetes in a production env
> ---------------------------------------
>
> Key: KYLIN-4447
> URL: https://issues.apache.org/jira/browse/KYLIN-4447
> Project: Kylin
> Issue Type: Sub-task
> Components: Integration
> Reporter: Xiaoxiang Yu
> Assignee: Xiaoxiang Yu
> Priority: Major
> Labels: kubernetes
> Fix For: v3.1.0
>
> Attachments: 79879570-73cfb200-8421-11ea-8ee3-108e903335a9.png,
> 79976447-31fd4500-84cf-11ea-83cb-52986ccc5173.png, BuildingSuccess.png
>
> Original Estimate: 120h
> Remaining Estimate: 120h
>
> h2. Kylin Cluster Deploy Diagram
>
> * ElasticSearch cluster is for log collection and analysis.
> * Hadoop cluster provided MR/Spark for cube building and HBase/HDFS for cube
> storage.
> * Kylin cluster is deployed on Kubernetes, including query servers and job
> servers.
> * Memcached cluster is both deployed on Kubernetes, it provided ability of
> tomcat session sharing and external query cache.
>
>
> !79976447-31fd4500-84cf-11ea-83cb-52986ccc5173.png!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
