search

sysulq / nginx-vts-exporter

Simple server that scrapes Nginx vts stats and exports them via HTTP for Prometheus consumption
https://sysulq.github.io/nginx-vts-exporter/
MIT License
645 stars 142 forks source link

nginx_upstream_response metric missing #57

Closed 123BLiN closed 6 years ago

123BLiN commented 6 years ago

Hello guys,

Am I missing something? Where is the main metric - nginx_upstream_response? As grafana dashboard has it preconfigured https://github.com/hnlq715/nginx-vts-exporter/blob/3b84c4ecab8ccb98dc59e6a8a00d76ab7d7b852f/dashboard/nginx-vts-exporter.json#L635 I may assume it was present earlier but now it is missing in the latest version (0.9.1), if this is expected can you advice the version where it is still present?

123BLiN commented 6 years ago

if this is expected can you advice the version where it is still present?

the version is 0.8 However there is one problem - for some reason instead of one metric that will have "backend" and "code" as tags we have two different metrics( So it is not possible to answer to the simple question - which service backend is having problems

bsushant-athena commented 6 years ago

see latest released version 0.9.1 https://github.com/hnlq715/nginx-vts-exporter/releases/tag/v0.9.1

123BLiN commented 6 years ago

https://github.com/hnlq715/nginx-vts-exporter/blob/master/nginx_vts_exporter.go#L196 from what I see "backend" tag is missing on the metric

sysulq commented 6 years ago

Could you please provide the detailed prometheus output here? Thanks. And I guess, maybe the grafana dashboard json file is outdated.

123BLiN commented 6 years ago

Hi! I'm sorry I propose to close this one, as the name of the topic is incorrect. I will create another one and possibly create pull request if I be able too. In short we need "backend" tags across all upstream metrics, because we need to be able differentiate upstream errors per host and also to track load on upstream hosts separately.