[Spark cluster standalone v2.4.0] - problems with reverse proxy functionnality regarding submitted applications in cluster mode and the spark history server ui
I have many spark clusters in standalone mode with 3 nodes each. One of them
is in HA with 3 masters and 3 workers and everything regarding the HA is
working fine. The second one is not in HA mode and we have one master and 3
In both of them, I have configured the reverse proxy for the UIs and
everything is working fine except two things:
1 - When I submit an application in cluster mode and the driver is launched
on a worker different from the one from which I do the submit command, the
access to the application UI through the master UI in reverse proxy is
broken and we have this error : *HTTP ERROR 502 \n Problem accessing
/myCustomProxyBase/proxy/app-20181203110740-0000. Reason: \n \tBad Gateway*
The workers UIs and this one which list the running executors
(/myCustomProxyBase/app/?appId=app-20181218104130-0006) work very fine.
However, when the driver is launched on the same worker from which the
submit command is executed, the access to the application through the master
works fine and others uris too (I mean the problem disappear).
2 - When the reverse proxy is setted, I can't access anymore to the history
server UI (many js and css errors) in both clusters.