Traefik should report traefik_backend_server_up 0 for unhealthy backends

[rinokadijk]

Do you want to request a feature or report a bug?

Bug

Did you try using a 1.7.x configuration for the version 2.0?

• Yes
• No

What did you do?

Scale an application that never becomes healthy in marathon. Traefik reports traefik_backend_server_up 1 for all backends

What did you expect to see?

According to the documentation:

# HELP traefik_backend_server_up Backend server is up, described by gauge value of 0 or 1.
# TYPE traefik_backend_server_up gauge
traefik_backend_server_up{backend="backend-my-service",url="http://xxx:yyy"} 1

For an unhealthy backend I would expect:

# HELP traefik_backend_server_up Backend server is up, described by gauge value of 0 or 1.
# TYPE traefik_backend_server_up gauge
traefik_backend_server_up{backend="backend-my-service",url="http://xxx:yyy"} 0

What did you see instead?

traefik_backend_server_up 1 for all (even unhealthy) backends

Output of traefik version: 1.7.9

1.7.9

What is your environment & configuration (arguments, toml, provider, platform, ...)?

debug = true

logLevel = "DEBUG"


[api]
entryPoint = "api"

[entryPoints]
    [entryPoints.http]
       address = ":80"
     [entryPoints.api]
       address = ":8090"  

[marathon]
    endpoint = "http://marathon:8080"
    domain = "marathon.localhost"
    exposedbydefault = true

[Ullaakut]

How do you know that your backends are unhealthy? What do you call unhealthy in your case?

If you could provide a way to reproduce your issue that'd be great, thanks.

[rinokadijk]

I would expect ANY application with a failing marathon healthcheck (https://docs.traefik.io/user-guide/marathon/#failure) to have:

traefik_backend_server_up{backend="backend-my-service",url="http://xxx:yyy"} 0

However, Traefik ALWAYS exposes:

traefik_backend_server_up{backend="backend-my-service",url="http://xxx:yyy"} 1

[martin-koerner]

E.g. for backends without any available Server there is obviously some check in the Web GUI - the backend is displayed red:

However, there is no prometheus metric exposing that this backend is even existing. I would expect a traefik_backend_server_up: 0 for this backend.

How do you know that your backends are unhealthy? What do you call unhealthy in your case?

If you could provide a way to reproduce your issue that'd be great, thanks.

[soleetions]

What's the status on this issue? I'm facing the same

[yuklia]

@jbdoumenjou, i have the same issue with unhealthy services.
i expect to see 0
traefik_backend_server_up{backend="backend-whoami-1"} 0
but in fact backend service just disappaered from metrics if service is actually down.

traefik version: V1.7.9