Limit parallelism
This commit is contained in:
parent
287424055e
commit
0ea086dc43
1 changed files with 4 additions and 0 deletions
|
@ -44,6 +44,8 @@ func scrapeVideos(startId string, fieldsToSend chan common.MapStr, done chan boo
|
|||
colly.Async(true),
|
||||
colly.MaxDepth(10),
|
||||
)
|
||||
videoCollector.Limit(&colly.LimitRule{DomainGlob: "*", Parallelism: 5})
|
||||
videoCollector.AllowURLRevisit = true
|
||||
|
||||
videoCollector.OnHTML("body", func(e *colly.HTMLElement) {
|
||||
url := e.Request.URL.String()
|
||||
|
@ -101,6 +103,8 @@ func (bt *Youtubebeat) Run(b *beat.Beat) error {
|
|||
ticker := time.NewTicker(bt.config.Period)
|
||||
for {
|
||||
select {
|
||||
case <-done:
|
||||
return nil
|
||||
case <-bt.done:
|
||||
return nil
|
||||
case <-ticker.C:
|
||||
|
|
Loading…
Reference in a new issue