Open surabujin opened 6 years ago
Scope:
We should be careful about logging the "processing time" per each command as it may flood the log. As a controllable solution we can use a set of metrics with a reporter which dumps the values either to the log or directly to the monitoring system. E.g. Timers + Slf4jReporter from Metrics
@rtretyak please share performance test results for this patch
I would like @surabujin to rebase the branch first, since the branch is way too old, many things changed since then
To be able to detect FL overload we need time metrics for commands being processed by FL. I.e we need to log how long each request have been processed (level info).