[SPARK-27996][WEBUI] Send HTTP redirect using the correct protocol when Spark History server is deployed behind the reverse proxy#27311
[SPARK-27996][WEBUI] Send HTTP redirect using the correct protocol when Spark History server is deployed behind the reverse proxy#27311ishikin wants to merge 3 commits intoapache:masterfrom
Conversation
|
ok to test |
|
Thank you for making a PR against |
|
Test build #117269 has finished for PR 27311 at commit
|
dongjoon-hyun
left a comment
There was a problem hiding this comment.
Hi, @ishikin .
The UT failure looks relevant one. Could you take a look at ApplicationCacheSuite?
org.apache.spark.deploy.history.ApplicationCacheSuite.redirect includes query params
…warded-Proto header
|
Hi @dongjoon-hyun, |
|
Test build #117363 has finished for PR 27311 at commit
|
core/src/test/scala/org/apache/spark/deploy/history/ApplicationCacheSuite.scala
Outdated
Show resolved
Hide resolved
core/src/test/scala/org/apache/spark/deploy/history/ApplicationCacheSuite.scala
Show resolved
Hide resolved
core/src/test/scala/org/apache/spark/deploy/history/ApplicationCacheSuite.scala
Show resolved
Hide resolved
|
Test build #117373 has finished for PR 27311 at commit
|
|
Retest this please. |
|
Test build #117390 has finished for PR 27311 at commit
|
|
Hi @dongjoon-hyun, |
|
@ishikin . Sorry. I also forgot this PR due to the other community work items( |
|
BTW, the following might be a risk at the other Spark resource managers, especially, YARN/MESOS.
|
|
We're closing this PR because it hasn't been updated in a while. This isn't a judgement on the merit of the PR in any way. It's just a way of keeping the PR queue manageable. |
What changes were proposed in this pull request?
This PR adds method
updateProtocolFromHeader(request: HttpServletRequest, url: String): Stringto JettyUtils.scala and calls this method from when Spark History Server is generating a redirect response.Why are the changes needed?
This change is needed to address bug SPARK-27996 in History Server. Please note that similar change is likely needed in Spark UI. The scope of this PR is only to address the problem in History Server.
Does this PR introduce any user-facing change?
Yes, the Spark History server deployed in Kubernetes now correctly opens the job page.
How was this patch tested?
The patch was tested by building and deploying Spark History Server to Kubernetes cluster and then observing the behavior of the History Server UI. The regression testing was not performed. I will appreciate is someone can can confirm the Spark History Server still works when deployed on regular infrastructure (e.g. not behind the reverse proxy).