prometheus_client: Internal Server Error does not log anything

[mmariani]

Bug report

With the Prometheus_client plugin, I get 500 Internal Server Error for some reason.
I don't have any information to debug it (even with agent.debug = true).

Relevant telegraf.conf:

[[outputs.prometheus_client]]
listen = ":9126"
expiration_interval = "60s"

System info:

Telegraf 1.3.0, Linux

Steps to reproduce:

wget http://localhost:9126/metrics
Connecting to localhost:9126 (127.0.0.1:9126)
wget: server returned error: HTTP/1.1 500 Internal Server Error

Expected behavior:

See something in stderr

Actual behavior:

Nothing gets logged

[danielnelson]

I wonder if it is caused by a particular input, can you try enabling only a single "system" input and see if you still receive a 500 error?

[mmariani]

It is indeed caused by docker input, but these metrics are exported to influxdb without issues.

[danielnelson]

This might be a duplicate of #2822, is there anything in the response body?

[mmariani]

Sorry I've been careless, I used wget at the beginning which did not print a response in case of error. With curl I had the complete response, and a lot of " label dimensions inconsistent with previously collected metrics"

[danielnelson]

I'm going to close this one, will track over on the other issue.