search

docker-flow / docker-flow-monitor

MIT License
87 stars 38 forks source link

Ability to define ScrapeInterval and ScrapeTimeout with labels #30

Closed ATLJLawrie closed 6 years ago

ATLJLawrie commented 6 years ago

Wasn't sure if this needed to go here or with the swarm-listener.

With the size of our cluster, using a global config of 30sec scrapes with 10 second timeouts, cadvisor can't seem to keep up with Prometheus. Even if we turn off all disable-able metrics via '-disable_metrics=tcp,udp,disk,network' the /metrics endpoint is about 2 MB. If I change my global to 60 / 30 things seem ok but it would be nice to do it more granularly.

What is your opinion or supporting something like com.df.scrapeInterval com.df.scrapeTimeout

vfarcic commented 6 years ago

Did you check GLOBAL__SCRAPE_INTERVAL from http://monitor.dockerflow.com/config/? I think that might be what you're looking for.

ATLJLawrie commented 6 years ago

Sorry for clarity. That is what I have done for now to resolve the issue. This was more to open an issue as a possible enhancement in which the individual target configs scrape interval and timeout could be set in the same manner with labels that port and endpoint are.

vfarcic commented 6 years ago

It's done and available in the release 18.02.10-50. Please try it out and let me know if it works as you expected.

ATLJLawrie commented 6 years ago

Finally got a chance to test this on 18.02.21-52 and it's working great.

militarpancho commented 5 years ago

Can anyone test if this feature is working properly?