libbeat: Stop autodiscover runners in a goroutine (elastic#7170) (ela…

…stic#7186)

Stopping runners can be a blocking operation, what can block the
autodiscover event loop. Run it in a goroutine so events can continue
being handled.

(cherry picked from commit eb58c3a)

CHANGELOG.asciidoc

@@ -121,6 +121,7 @@ https://github.com/elastic/beats/compare/v6.2.3...v6.3.0[View commits]
 - Ensure Kubernetes labels/annotations don't break mapping {pull}6490[6490]
 - Ensure that the dashboard zip files can't contain files outside of the kibana directory. {pull}6921[6921]
 - Fix map overwrite panics by cloning shared structs before doing the update. {pull}6947[6947]
+- Fix delays on autodiscovery events handling caused by blocking runner stops. {pull}7170[7170]
 
 *Auditbeat*
 

libbeat/autodiscover/autodiscover.go

@@ -174,8 +174,12 @@ func (a *Autodiscover) handleStop(event bus.Event) {
 		}
 
 		if runner := a.runners.Get(hash); runner != nil {
+			// Stop can block, we run it asyncrhonously to avoid blocking
+			// the whole events loop. The runner hash is removed in any case
+			// so an equivalent configuration can be added again while this
+			// one stops, and a duplicated event don't try to stop it twice.
 			logp.Info("Autodiscover stopping runner: %s", runner)
-			runner.Stop()
+			go runner.Stop()
 			a.runners.Remove(hash)
 		} else {
 			logp.Debug(debugK, "Runner not found for stopping: %s", hash)