[
https://issues.apache.org/jira/browse/SPARK-11147?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14962043#comment-14962043
]
Sebastian YEPES FERNANDEZ commented on SPARK-11147:
---------------------------------------------------
Ok I have found the source of my problem. For a bit of background, all of
our nodes are multihomed: one public 1GB NIC only for admin access and a
second internal 10GB NIC dedicated for all the cluster traffic
(yarn,hdfs,spark...)
Last night after looking at the source code, I thought it could actually be a
networking issue. So I tried several settings and found a solution.
Solution:
{code}
# Globally export this variable on all the nodes with there corresponding
internal IP
echo "export SPARK_LOCAL_IP=192.168.1.x" >/etc/profile
# Restart all the YARN services
{code}
After making these changes now when I submit a job in cluster mode I can access
the SparkIU.
Every thing now works, but there is still is strange thing in the UI. As Steve
mentioned the "webproxy" settings are showing the incorrect app ID's (see
attachment)
> HTTP 500 if try to access Spark UI in yarn-cluster
> --------------------------------------------------
>
> Key: SPARK-11147
> URL: https://issues.apache.org/jira/browse/SPARK-11147
> Project: Spark
> Issue Type: Bug
> Components: Web UI, YARN
> Affects Versions: 1.5.1
> Environment: HDP: 2.3.2.0-2950 (Hadoop 2.7.1.2.3.2.0-2950)
> Spark: 1.5.x (c27e1904)
> Reporter: Sebastian YEPES FERNANDEZ
>
> Hello,
> I am facing a similar issue as described in SPARK-5837, but is my case the
> SparkUI only work in "yarn-client" mode. If a run the same job using
> "yarn-cluster" I get the HTTP 500 error:
> {code}
> HTTP ERROR 500
> Problem accessing /proxy/application_1444297190346_0085/. Reason:
> Connection to http://XX.XX.XX.XX:55827 refused
> Caused by:
> org.apache.http.conn.HttpHostConnectException: Connection to
> http://XX.XX.XX.XX:55827 refused
> at
> org.apache.http.impl.conn.DefaultClientConnectionOperator.openConnection(DefaultClientConnectionOperator.java:190)
> at
> org.apache.http.impl.conn.ManagedClientConnectionImpl.open(ManagedClientConnectionImpl.java:294)
> at
> org.apache.http.impl.client.DefaultRequestDirector.tryConnect(DefaultRequestDirector.java:643)
> at
> org.apache.http.impl.client.DefaultRequestDirector.execute(DefaultRequestDirector.java:479)
> at
> org.apache.http.impl.client.AbstractHttpClient.execute(AbstractHttpClient.java:906)
> {code}
> I have verified that the UI port "55827" is actually Listening on the worker
> node, I can even run a "curl http://XX.XX.XX.XX:55827"; and it redirects me to
> another URL: http://YY.YY.YY.YY:8088/proxy/application_1444297190346_0082
> The strange thing is the its redirecting me to the app "_0082" and not the
> actually running job "_0085"
> Does anyone have any suggestions on what could be causing this issue?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
