You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by GitBox <gi...@apache.org> on 2022/03/28 15:55:04 UTC

[GitHub] [hadoop] kiran-maturi commented on pull request #3445: HADOOP-15566 Opentelemetry changes using java agent

kiran-maturi commented on pull request #3445:
URL: https://github.com/apache/hadoop/pull/3445#issuecomment-1080825440


   
   > * wire marshalling in protobuf: efficiency and compatibility
   My thought for having a map was, the trace_id, parent_id, trace_flags, trace_state have fixed lengths the serialization and deserialization is still in progress for binary format. It might be good to wait for sometime based on this 
   https://w3c.github.io/trace-context-binary/. 
   
   > * what our story about dependencies are.
   Currently we need the agent jar only at runtime and its shaded already. We need to have in seperate path other than the common which will not be picked by default (ex $HADOOP_HOME/share/hadoop/trace). probably somewhere outside the common. Opentelemetry APIs are light . The agent jar will only be picked from env when enabled.  
   # export OPENTELEMETRY_JAVAAGENT_PATH="$(find $HADOOP_HOME/share/hadoop/tools/lib/ -name opentelemetry-javaagent*)"
   
   > for dependencies we either embrace and mandate a new jar on the cp for iPC (easiest in our code) or do some reflection games to downgrade if it is not on the classpath. which, given telemetry isn't normally a critical feature, could be justified.
   
   I am not sure if we need this
   
   > how much extra pain in terms of transient dependencies would it be for us to always make the jar a dependency of hadoop common?
   
   We can keep it common the whole size was close to 28MB and its shaded.
   
   Please share your thoughts
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org