VTS Upstream groups not working with --enable-dynamic-configuration

[valeriano-manassero]

Using --enable-dynamic-configuration causes Nginx Vhost Traffic Status on /nginx_status to show only one upstream group:

upstream_balancer

This group metrics are never updated too.
This is due of LUA taking care of load balancing of the only upstream group declared in nginx.conf.
But this behavior completely eliminates any backend metric.

I already tried to workaround this but without luck. Have you any ideas on how to tackle this?

[aledbf]

This group metrics are never updated too.

We are aware of this limitation

I already tried to workaround this but without luck. Have you any ideas on how to tackle this?

Not yet. I think we need to start porting some features like this one and Prometheus to LUA land using a custom module using go as the backend of the module to avoid the limitations of LUA dictionaries for this kind of tasks

[aledbf]

@ElvinEfendi ping

[ElvinEfendi]

We have started looking into this in the context of monitoring ingress-nginx in general. My colleague @fmejia97 is working on this. The idea is first to document the existing monitoring options, then document what's desired and what's missing and then implement the missing ones.

[mikksoone]

Any update on this? Is the plan to fake it for VTS or use custom monitoring solution that provides similar data via prometheus etc?

[ElvinEfendi]

@mikksoone we are expecting to have something by the end of this month. The plan is to implement something similar to what VTS provides in Lua.

[valeriano-manassero]

There are some progress on this?
Thank you.

[aledbf]

@valeriano-manassero yes, please check #2608

[aledbf]

@valeriano-manassero just in case, please don't use this change yet, we need to clean up some things.