afedulov commented on a change in pull request #11195: [FLINK-16222][runtime]
Use plugins mechanism for initializing MetricReporters
URL: https://github.com/apache/flink/pull/11195#discussion_r393839754
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/metrics/ReporterSetup.java
##########
@@ -120,55 +124,37 @@ public static ReporterSetup forReporter(String
reporterName, MetricConfig metric
}
private static ReporterSetup createReporterSetup(String reporterName,
MetricConfig metricConfig, MetricReporter reporter) {
- LOG.info("Configuring {} with {}.", reporterName, metricConfig);
+ LOG.debug("Configuring {} with {}.", reporterName,
metricConfig);
Review comment:
Ping @AHeise @zentol - could we agree on something here?
I am always "pro extensive logging", but this could be professional
deformation. Being able to "on-demand" see what is going on is very valuable
for production systems. We could declare somewhere for Flink in general that if
you choose to run with debug log level, some potentially sensitive information
could leak into logs. My arguments are:
1) If someone has uncontrolled access to the log files on your machine in
production, content of this file is probably not the biggest of your problems.
2) Running with debug level is not a "normal" scenario - this is intended
for hands on investigation of issues. Log level for potentially compromisable
external systems could be explicitly set to trace in such cases.
3) We have been "leaking" this data in the current versions with info (!)
level without much concern
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services