Metric rename in prometheus input and multiple prometheus outputs

[zagy]

In #2950 the renaming was generally fixed (so that haproxy_up doesn't become haproxy_up_gauge). When there is more than one Prometheus output, only the last yields the desired result.

Relevant telegraf.conf:

[inputs]

[[inputs.prometheus]]
urls = ["http://localhost:9127/metrics"]

[outputs]

[[outputs.prometheus_client]]
listen = "172.22.50.75:2000"

[[outputs.prometheus_client]]
listen = "172.22.50.75:2001"

Note: multiple outputs on different ports is pretty pointless. The actual use case is IPv4+IPv6 on a certain interface.

System info:

• Telegraf 1.5.3

Steps to reproduce:

1. Setup https://github.com/prometheus/haproxy_exporter
2. Point telegraf to haproxy_export (prometheus input)
3. Setup telegraf output prometheus_client
4. See that metrics are renamed after they come out of telegraf:

First output:

$ curl http://172.22.50.75:2000/metrics -s | grep haproxy_up
# HELP haproxy_up_gauge Telegraf collected metric
# TYPE haproxy_up_gauge untyped
haproxy_up_gauge{host="apo00",location="rzob",platform="nixos",resource_group="apo",url="http://localhost:9127/metrics"} 1

That was wrong. Next output is better:

$  curl http://172.22.50.75:2001/metrics -s | grep haproxy_up
# HELP haproxy_up Telegraf collected metric
# TYPE haproxy_up gauge
haproxy_up{host="apo00",location="rzob",platform="nixos",resource_group="apo",url="http://localhost:9127/metrics"} 1

Expected behavior:

• No rename of metric
• Keep type

Actual behavior:

• Metric is being renamed to *_ (e.g. haproxy_up becomes haproxy_up_gauge)
• Type becomes untyped

[danielnelson]

Thanks for the report. I'm not sure the cause, but this appears to be fixed in the latest release. Since I am no longer providing updates to the 1.5 line you will need to update.

[zagy]

No problem with updating. Looking through the changes didn't reveal a fix for me, but if it's fixed that's great. Sorry for the noise.

[danielnelson]

No problem, yeah I'm not sure what fixed it either.