Improve page loading error handling (#58)

Author: man90es

README.md

@@ -62,6 +62,10 @@ Available flags:
 	- Specifies Redis server's address
 	- Type: string
 	- Default value: none
+- `-scraperfailurepause`
+	- Amount of time in seconds to wait after a failed task to idle (0 for immediate retry, negative for pause until end of minute)
+	- Type: integer
+	- Default value: `-1`
 
 You can use them like this:
 ```bash

handlers/getStatus.go

@@ -12,7 +12,7 @@ import (
 )
 
 var initTime = time.Now()
-var version = "1.17.0"
+var version = "1.18.0"
 
 func getStatus(w http.ResponseWriter, r *http.Request) {
 	json.NewEncoder(w).Encode(map[string]interface{}{

main.go

@@ -26,6 +26,7 @@ func main() {
 	flagProxyReloadWebhook := flag.String("proxyreloadwebhook", "", "Webhook address to request proxy reload")
 	flagRateLimit := flag.Uint64("ratelimit", 512, "Maximum number of requests per minute per IP")
 	flagRedis := flag.String("redis", "", "Redis connection string")
+	flagScraperFailurePause := flag.Int("scraperfailurepause", -1, "Amount of time in seconds to wait after a failed task to idle")
 	flagTaskRetries := flag.Uint("taskretries", 3, "Number of retries for a scraping task")
 	flagVerbose := flag.Bool("verbose", false, "Print out additional logs into stdout")
 	flag.Parse()
@@ -58,6 +59,7 @@ func main() {
 	viper.Set("proxyreloadwebhook", *flagProxyReloadWebhook)
 	viper.Set("ratelimit", int64(*flagRateLimit))
 	viper.Set("redis", *flagRedis)
+	viper.Set("scraperfailurepause", time.Duration(*flagScraperFailurePause)*time.Second)
 	viper.Set("taskretries", int(*flagTaskRetries))
 	viper.Set("verbose", *flagVerbose)
 

scraper/handleTaskError.go

@@ -0,0 +1,44 @@
+package scraper
+
+import (
+	"bdo-rest-api/logger"
+	"bdo-rest-api/utils"
+	"fmt"
+	"strconv"
+	"time"
+
+	"github.com/gocolly/colly/v2"
+	"github.com/spf13/viper"
+)
+
+func handleTaskError(r *colly.Request, imperva bool, err error) {
+	taskRetries, _ := strconv.Atoi(r.Ctx.Get("taskRetries"))
+
+	if imperva {
+		logger.Error(fmt.Sprintf("Hit Imperva while loading %v, retries: %v", r.URL, taskRetries))
+	} else {
+		logger.Error(fmt.Sprintf("Error occured while loading %v: %v, retries: %v", r.URL, err, taskRetries))
+	}
+
+	if proxyReloadWebhook := viper.GetString("proxyreloadwebhook"); proxyReloadWebhook != "" {
+		utils.SendDummyRequest(proxyReloadWebhook)
+	}
+
+	if scraperFailurePause := viper.GetDuration("scraperfailurepause"); scraperFailurePause >= 0 {
+		taskQueue.Pause(scraperFailurePause)
+	} else {
+		taskQueue.Pause(time.Duration(60-time.Now().Second()) * time.Second)
+	}
+
+	taskQueue.ConfirmTaskCompletion(r.Ctx.Get("taskClient"), r.Ctx.Get("taskHash"))
+
+	if taskRetries < viper.GetInt("taskretries") {
+		taskRegion := r.Ctx.Get("taskRegion")
+		taskType := r.Ctx.Get("taskType")
+		taskQueue.AddTask(r.Ctx.Get("taskClient"), r.Ctx.Get("taskHash"), utils.BuildRequest(r.URL.String(), map[string]string{
+			"taskRegion":  taskRegion,
+			"taskRetries": strconv.Itoa(taskRetries + 1),
+			"taskType":    taskType,
+		}))
+	}
+}

scraper/scraper.go

@@ -58,8 +58,7 @@ func InitScraper() {
 	})
 
 	scraper.OnError(func(r *colly.Response, err error) {
-		logger.Error(fmt.Sprintf("Error occured while loading %v: %v", r.Request.URL, err))
-		taskQueue.ConfirmTaskCompletion(r.Ctx.Get("taskClient"), r.Ctx.Get("taskHash"))
+		handleTaskError(r.Request, false, err)
 	})
 
 	scraper.OnResponse(func(r *colly.Response) {
@@ -80,24 +79,7 @@ func InitScraper() {
 		})
 
 		if imperva {
-			taskRetries, _ := strconv.Atoi(body.Request.Ctx.Get("taskRetries"))
-			logger.Error(fmt.Sprintf("Hit Imperva while loading %v, retries: %v", body.Request.URL.String(), taskRetries))
-			if proxyReloadWebhook := viper.GetString("proxyreloadwebhook"); proxyReloadWebhook != "" {
-				utils.SendDummyRequest(proxyReloadWebhook)
-				taskQueue.Pause(time.Second * 5)
-			} else {
-				taskQueue.Pause(time.Duration(60-time.Now().Second()) * time.Second)
-			}
-			taskQueue.ConfirmTaskCompletion(taskClient, taskHash)
-
-			if taskRetries < viper.GetInt("taskretries") {
-				taskQueue.AddTask(taskClient, taskHash, utils.BuildRequest(body.Request.URL.String(), map[string]string{
-					"taskRegion":  taskRegion,
-					"taskRetries": strconv.Itoa(taskRetries + 1),
-					"taskType":    taskType,
-				}))
-			}
-
+			handleTaskError(body.Request, true, nil)
 			return
 		}