HeartSaVioR edited a comment on issue #23260: [SPARK-26311][YARN] New feature: 
custom log URL for stdout/stderr
URL: https://github.com/apache/spark/pull/23260#issuecomment-455328233
 
 
   @squito Yes, exactly. Nice summarization.
   
   >  this includes when the SHS is showing "inprogress" applications (this 
could be changed I think)
   
   IMHO there're some cases we still want to get custom log URL while the 
status of application is "in progress", because according to the logic on 
determining whether the application is finished, app could be shown as 
"inprogress" when the app terminated unexpectedly.
   
   
https://github.com/apache/spark/blob/0b3abef1950f486001160ec578e4f628c199eeb4/core/src/main/scala/org/apache/spark/deploy/history/HistoryPage.scala#L98-L100
   
   In this case, showing origin log URLs may not work but showing custom log 
URLs would work if the external log service gathers executor logs continuously 
rather than gathering log when app is finished - I expect the behavior of 
external log service as former not latter. Even if app is still running, app UI 
will still provide origin log URLs so we can get it from there instead of SHS.
   
   What I agreed to make change on only SHS was my prev. patch was making log 
urls being static again which is fragile if there's a change on external log 
service. Now we just require end users to modify the configuration and restart 
SHS (unless there's a cache mechanism to avoid re-reading events when 
restarting SHS.) IMHO, showing custom log URLs even for apps which are shown as 
"inprogress" still makes sense to me. 
   
   Btw, I didn't add variables for other resource managers as they don't 
provide log URLs for now. (If possible we may be better to address it as well 
via separate PRs.)

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to