Hi Kumar, MDC is stored in Thead Local variable. When you are using a thead pool: the calling thread A pushes an event in a wait queue, and then (sometimes later), the called thread B pops the event from the queue and runs it.
As a result, 1) the MDC is not propagated from thread A to thread B. 2) the MDC of thread B may reused for all events consumed from the queue: this can cause wrong MDC. To solver these problems, you should: 1) MDC not propagated: Thread A copies the MDC info in the event, Thread B sets MDC from info in the event. 2) MDC is reused: Thread B clears the whole MDC before running a new event. This is not a problem with SLF4J but with thread locals in general. Gérald Le jeu. 27 sept. 2018 à 12:48, kumar saurabh <sabs.bl...@gmail.com> a écrit : > Hi, > > I am working on microservices stack and using correlation id (Unique > request id) generated via nginx to log into each microservice logs. I am > using MDC for extracting this from request header in OncePerRequestFilter > in java. > > One of the microservice then use java thread executor framework and then > call other microservices. > > Can using MDC with executor (java thread pool) mess up due to thread re > use? Is it possible that I get either empty MDC context or incorrect > context due to thread reuse? > > I am referring below link where one such scenario is explained: > http://ashtonkemerling.com/blog/2017/09/01/mdc-and-threadpools/ > > A response is appreciated! Thanks > > -- > Regards, > Kumar Saurabh > _______________________________________________ > slf4j-user mailing list > slf4j-user@qos.ch > http://mailman.qos.ch/mailman/listinfo/slf4j-user
_______________________________________________ slf4j-user mailing list slf4j-user@qos.ch http://mailman.qos.ch/mailman/listinfo/slf4j-user
