Metrics are collected and merged in http.prometheus, no need to be run
on every worker.
- kr_request::qsource.tcp renamed to ::qsource.flags.tcp
- kr_request::has_tls renamed to ::qsource.flags.tls
+Bugfixes
+--------
+- http module: only run prometheus in parent process if using --forks=N,
+ as the submodule collects metrics from all sub-processes as well.
+
Knot Resolver 3.1.0 (2018-11-02)
================================
-- @function Init module
function M.init()
- worker.coroutine(prometheus.init)
+ -- collect and merge metrics only on leader
+ if worker.id == 0 then
+ worker.coroutine(prometheus.init)
+ end
end
-- @function Cleanup module