Add scrapeTimeout as global configurable parameter #3250
Conversation
|
Works fine on my ARM cluster: Jsonnet:
Config:
|
|
lgtm on green |
|
Tests seem flaky. Fails seems not related to the change as I looked into the PR test runs and lots fail around |
|
hmm I kicked the jobs a couple of times, I think there is actually something not working |
|
It seems be consistently failing on |
pkg/prometheus/promcfg.go
Outdated
| @@ -177,6 +177,11 @@ func (cg *configGenerator) generateConfig( | |||
| scrapeInterval = p.Spec.ScrapeInterval | |||
| } | |||
|
|
|||
| scrapeTimeout := "10s" | |||
There was a problem hiding this comment.
Whats the reason for the 10seconds here, might have missed something?
There was a problem hiding this comment.
It's the default timeout. Even changing to 30s the same test fails. I don't think it is related to the change.
|
I've reverted the timeout to 10s (the default Prometheus scrape_timeout) and added a 60s timeout for the failing test. |
Add global configurable scrapeTimeout parameter to allow monitoring targets on clusters consisting of slower hosts like Raspberry Pi and many ARM boards used for labs. Signed-off-by: Carlos de Paula <me@carlosedp.com>
|
Now we're good to go. Better and cleaner implementation, also Travis is green! |
Add global configurable scrapeTimeout parameter to allow monitoring
targets on clusters consisting of slower hosts like Raspberry Pi and
many ARM boards used for labs.
Many times the 10s default timeout is not enough for targets like kube-controller or kube-scheduler so some targets get a "Context deadline exceeded" error.
Signed-off-by: Carlos de Paula me@carlosedp.com