Callback duration metric

chronos/asyncloop.nim

@@ -15,6 +15,12 @@ import std/[tables, strutils, heapqueue, deques]
 import stew/results
 import "."/[config, futures, osdefs, oserrno, osutils, timer]
 
+when chronosEnableCallbackDurationMetric:
+  import std/monotimes
+  import pkg/metrics
+
+  declareGauge(chronosCallbackDuration, "chronos callback duration")
+
 export Port
 export futures, timer, results
 
@@ -260,8 +266,15 @@ template processCallbacks(loop: untyped) =
     if isSentinel(callable):
       break
     if not(isNil(callable.function)):
+      when chronosEnableCallbackDurationMetric:
+        let startTime = getMonoTime().ticks
+
       callable.function(callable.udata)
 
+      when chronosEnableCallbackDurationMetric:
+        let durationUs = (getMonoTime().ticks - startTime) div 1000
+        chronosCallbackDuration.set(durationUs)
+
 proc raiseAsDefect*(exc: ref Exception, msg: string) {.noreturn, noinline.} =
   # Reraise an exception as a Defect, where it's unexpected and can't be handled
   # We include the stack trace in the message because otherwise, it's easily

chronos/config.nim

@@ -70,6 +70,10 @@ when (NimMajor, NimMinor) >= (1, 4):
         ""
       ## OS polling engine type which is going to be used by chronos.
 
+    chronosEnableCallbackDurationMetric* {.booldefine.} = defined(chronosEnableCallbackDurationMetric)
+      ## At the cost of some performance, produce a 'chronosCallbackDuration' metric.
+      ## Useful for detecting application stalling/blocking.
+
 else:
   # 1.2 doesn't support `booldefine` in `when` properly
   const