Could it be that you are submitting the job in attached mode, i.e., without -d
parameter? In the "job cluster attached mode", we actually start a Flink
session cluster (and stop it again from the CLI) . Therefore, in attached
mode, the config option "yarn.per-job-cluster.include-user-jar" is effectively
ignored. If you submit with -d, a "true job cluster" is started, and the user
jar should be added to the system classpath. Alternatively, if the detached
mode is not an option for you, you could add a jar with your custom logger
implementation to the flink /lib directory.
If the behavior in Flink 1.3 is indeed different, then I would consider this a
regression. Can you open a jira issue for that?