Many errors when getting /metrics

[valvin1]

Hello,

I'm using version v0.5.0 in its docker version. I collect metrics coming from rhel6 and 7.

collectd version on rhel6 is 4.10.9-5
collectd version on rhel7 is 5.8.1-1

I had collectd_exporter working with only rhel6 for some times it looks rhel7 bring this regression.

Here is the one of the 350 errors I get when I call /metrics :

An error has occurred while serving metrics:

350 error(s) occurred:
[...]
* collected metric collectd_interface_if_errors_1_total label:<name:"instance" value:"myserver" > label:<name:"interface" value:"eth1" > counter:<value:0 >  has help "Collectd exporter: 'interface' Type: 'if_errors' Dstype: 'api.Counter' Dsname: '1'" but should have "Collectd exporter: 'interface' Type: 'if_errors' Dstype: 'api.Derive' Dsname: '1'"

it appears on many different metrics.

Collectd is configured using network plugin :

LoadPlugin network
<Plugin network>
  Server "collectd-exporter-server" "25826"
</Plugin>

I'll test to remove rhel7 server and see if it works back. I've already tries to destroy container. I was first on latest version and switch to named version v0.5.0

Does this issue is known ? Should i switch on write_http plugin for rhel7?

[valvin1]

I confirm issue is coming from server running collectd v5.8.1 on rhel7. I've desactivated it and it works back.

[eero-t]

Collectd v4.10.9 is from 2013, and 5.8.1 is from 2018, i.e. both rather old. There's also new v0.6.0 release of this project: https://github.com/prometheus/collectd_exporter/releases/tag/v0.6.0

Does this issue is known ? Should i switch on write_http plugin for rhel7?

I have tried neither collectd_exporter (i.e. pushing metrics with this project), nor write_http (pulling metrics from collectd). I'm using write_prometheus plugin from collectd, and that works fine for me...