search

flightstats / hub

fault tolerant, highly available service for data storage and distribution
http://www.flightstats.com
MIT License
103 stars 35 forks source link

Use timing method #1161

Closed Paul-Hess closed 5 years ago

Paul-Hess commented 5 years ago

https://jira.flightglobal.io/browse/PDXDDT-134

edited with updated datadog link:

https://app.datadoghq.com/dashboard/wsq-d9n-26w/hubs-by-cluster?tile_size=l&page=0&is_auto=false&from_ts=1555621395000&to_ts=1555622295000&live=true&tpl_var_cluster=dls.dev&fullscreen_widget=3543467067562178

lkemmerer commented 5 years ago

Dunno if it'd make sense to keep both metrics around until we decide which one makes more sense, but if you feel like this is definitely the right metric, I'm cool with that, too

lkemmerer commented 5 years ago

...nevermind, for some reason I forgot you can test this in dev, despite the link. :P

Paul-Hess commented 5 years ago

the link to the Datadog graph is updated and I think displays the intended use of both metrics together. To get a similar graphing I think the hub would have to do more work that I am fine with handing off to the <influxdb|datadog> query

lkemmerer commented 5 years ago

Makes sense. I am a little weirded out that the time spent in the queue is 0, but it might just be that it's a relatively low-volume dev cluster.

Paul-Hess commented 5 years ago

I believe that the oddly low metrics are because it is deployed on dls dev