Add support for timeout offset (#647)

* Add support for timout offset

Allow the MongoDB timeout to be configured relative to the Prometheus
scrape timeout. This ensures the Prometheus scrape will get a response even
if the exporter cannot connect to MongoDB.

* Add check to ensure TimeoutOffset is greater than 0

* Empty commit

REFERENCE.md

@@ -14,6 +14,7 @@
 |--web.listen-address|Address to listen on for web interface and telemetry|--web.listen-address=":9216"|
 |--web.telemetry-path|Metrics expose path|--web.telemetry-path="/metrics"|
 |--web.config|Path to the file having Prometheus TLS config for basic auth|--web.config=STRING|
+|--web.timeout-offset|Offset to subtract from the timeout in seconds|--web.timeout-offset=1|
 |--log.level|Only log messages with the given severity or above. Valid levels: [debug, info, warn, error, fatal]|--log.level="error"|
 |--collector.diagnosticdata|Enable collecting metrics from getDiagnosticData|
 |--collector.replicasetstatus|Enable collecting metrics from replSetGetStatus|
@@ -27,4 +28,4 @@
 |--collector.profile-time-ts=30|Set time for scrape slow queries| This interval must be synchronized with the Prometheus scrape interval|
 |--collector.profile|Enable collecting metrics from profile|
 |--metrics.overridedescendingindex| Enable descending index name override to replace -1 with _DESC ||
-|--version|Show version and exit|
+|--version|Show version and exit|

exporter/exporter.go

@@ -56,6 +56,7 @@ type Opts struct {
 	DiscoveringMode        bool
 	GlobalConnPool         bool
 	ProfileTimeTS          int
+	TimeoutOffset          int
 
 	CollectAll               bool
 	EnableDBStats            bool
@@ -271,6 +272,7 @@ func (e *Exporter) Handler() http.Handler {
 		if err != nil {
 			seconds = 10
 		}
+		seconds -= e.opts.TimeoutOffset
 
 		var client *mongo.Client
 		ctx, cancel := context.WithTimeout(r.Context(), time.Duration(seconds)*time.Second)

main.go

@@ -44,6 +44,7 @@ type GlobalFlags struct {
 	WebListenAddress      string   `name:"web.listen-address" help:"Address to listen on for web interface and telemetry" default:":9216"`
 	WebTelemetryPath      string   `name:"web.telemetry-path" help:"Metrics expose path" default:"/metrics"`
 	TLSConfigPath         string   `name:"web.config" help:"Path to the file having Prometheus TLS config for basic auth"`
+	TimeoutOffset         int      `name:"web.timeout-offset" help:"Offset to subtract from the request timeout in seconds" default:"1"`
 	LogLevel              string   `name:"log.level" help:"Only log messages with the given severity or above. Valid levels: [debug, info, warn, error, fatal]" enum:"debug,info,warn,error,fatal" default:"error"`
 	ConnectTimeoutMS      int      `name:"mongodb.connect-timeout-ms" help:"Connection timeout in milliseconds" default:"5000"`
 
@@ -112,6 +113,11 @@ func main() {
 		ctx.Fatalf("No MongoDB hosts were specified. You must specify the host(s) with the --mongodb.uri command argument or the MONGODB_URI environment variable")
 	}
 
+	if opts.TimeoutOffset <= 0 {
+		log.Warn("Timeout offset needs to be greater than \"0\", falling back to \"1\". You can specify the timout offset with --web.timeout-offset command argument")
+		opts.TimeoutOffset = 1
+	}
+
 	serverOpts := &exporter.ServerOpts{
 		Path:             opts.WebTelemetryPath,
 		MultiTargetPath:  "/scrape",
@@ -135,6 +141,7 @@ func buildExporter(opts GlobalFlags, uri string, log *logrus.Logger) *exporter.E
 		GlobalConnPool:        opts.GlobalConnPool,
 		DirectConnect:         opts.DirectConnect,
 		ConnectTimeoutMS:      opts.ConnectTimeoutMS,
+		TimeoutOffset:         opts.TimeoutOffset,
 
 		EnableDiagnosticData:     opts.EnableDiagnosticData,
 		EnableReplicasetStatus:   opts.EnableReplicasetStatus,