I figured it all out after this: http://apache-spark-user-list.1001560.n3.nabble.com/WebUI-on-yarn-through-ssh-tunnel-affected-by-AmIpfilter-td21540.html
The short is that I needed to set SPARK_PUBLIC_DNS (not DNS_HOME) = ec2_publicdns then the YARN proxy gets in the way, so I needed to go to: http://ec2_publicdns:20888/proxy/applicationid/jobs (9046 is the older emr port) or, as Jonathan said, the spark history server works once a job is completed. On Tue, Aug 25, 2015 at 5:26 PM, Justin Pihony <justin.pih...@gmail.com> wrote: > OK, I figured the horrid look also....the href of all of the styles is > prefixed with the proxy data....so, ultimately if I can fix the proxy > issues with the links, then I can fix the look also.... > > On Tue, Aug 25, 2015 at 5:17 PM, Justin Pihony <justin.pih...@gmail.com> > wrote: > >> SUCCESS! I set SPARK_DNS_HOME=ec2_publicdns, which makes it available to >> access the spark ui directly. The application proxy was still getting in >> the way by the way it creates the URL, so I manually filled in the >> /stage?id=#&attempt=# and that worked....I'm still having trouble with the >> css as the UI looks horrid....but I'll tackle that next :) >> >> On Tue, Aug 25, 2015 at 4:31 PM, Justin Pihony <justin.pih...@gmail.com> >> wrote: >> >>> Thanks. I just tried and still am having trouble. It seems to still be >>> using the private address even if I try going through the resource manager. >>> >>> On Tue, Aug 25, 2015 at 12:34 PM, Kelly, Jonathan <jonat...@amazon.com> >>> wrote: >>> >>>> I'm not sure why the UI appears broken like that either and haven't >>>> investigated it myself yet, but if you instead go to the YARN >>>> ResourceManager UI (port 8088 if you are using emr-4.x; port 9026 for >>>> 3.x, >>>> I believe), then you should be able to click on the ApplicationMaster >>>> link >>>> (or the History link for completed applications) to get to the Spark UI >>>> from there. The ApplicationMaster link will use the YARN Proxy Service >>>> (port 20888 on emr-4.x; not sure about 3.x) to proxy through the Spark >>>> application's UI, regardless of what port it's running on. For completed >>>> applications, the History link will send you directly to the Spark >>>> History >>>> Server UI on port 18080. Hope that helps! >>>> >>>> ~ Jonathan >>>> >>>> >>>> >>>> >>>> On 8/24/15, 10:51 PM, "Justin Pihony" <justin.pih...@gmail.com> wrote: >>>> >>>> >I am using the steps from this article >>>> ><https://aws.amazon.com/articles/Elastic-MapReduce/4926593393724923> >>>> to >>>> >get spark up and running on EMR through yarn. Once up and running I >>>> ssh in >>>> >and cd to the spark bin and run spark-shell --master yarn. Once this >>>> spins >>>> >up I can see that the UI is started at the internal ip of 4040. If I >>>> hit >>>> >the >>>> >public dns at 4040 with dynamic port tunneling and foxyproxy then I >>>> get a >>>> >crude UI (css seems broken), however the proxy continuously redirects >>>> me >>>> >to >>>> >the main page, so I cannot drill into anything. So, I tried static >>>> >tunneling, but can't seem to get through. >>>> > >>>> >So, how can I access the spark UI when running a spark shell in AWS >>>> yarn? >>>> > >>>> > >>>> > >>>> >-- >>>> >View this message in context: >>>> > >>>> http://apache-spark-user-list.1001560.n3.nabble.com/How-to-access-Spark-UI >>>> >-through-AWS-tp24436.html >>>> >Sent from the Apache Spark User List mailing list archive at >>>> Nabble.com. >>>> > >>>> >--------------------------------------------------------------------- >>>> >To unsubscribe, e-mail: user-unsubscr...@spark.apache.org >>>> >For additional commands, e-mail: user-h...@spark.apache.org >>>> > >>>> >>>> >>> >> >