nuvo/emq_exporter

Simple server that scrapes EMQ metrics and exporters them via HTTP for Prometheus consumption

emq_exporter's Issues

Instructions on running emq_exporter in docker

I run emq_exporter:0.4.1 by docker, I get error follow, how to run emq_exporter by docker?

[root@localhost ~]# docker run -p 9540:9540 nuvo/emq_exporter:v0.3.1 ---emq.uri "http://localhost:8080"

Unable to find image 'nuvo/emq_exporter:v0.3.1' locally
v0.3.1: Pulling from nuvo/emq_exporter
57c14dd66db0: Already exists 
be921ee89692: Already exists 
f85a6778cbc5: Pull complete 
Digest: sha256:61f8ebb871f9e49a6a09c8bd4d8bf2ef3ac3316b14972efb3c091964d0300fb1
Status: Downloaded newer image for nuvo/emq_exporter:v0.3.1
emq_exporter: error: unknown long flag '---emq.uri', try --help

Switch to go modules

Support V3 API for emqx

V3 changed some of the metrics API paths.

We currently have the plumbing to support this, needs implementation and validation.

Failed to get metrics from http://127.0.0.1:8080

I got error:
[root@localhost emq_exporter-0.2.2-linux]# ./emq_exporter --emq.uri="http://127.0.0.1:8080" --emq.node="[email protected]" --emq.api-version="v3" INFO[0000] Starting emq_exporter source="emq_exporter.go:274"
INFO[0000] Version v0.2.2 (git-5200959) source="emq_exporter.go:275"
INFO[0000] Listening on :9505 source="emq_exporter.go:284"
WARN[0005] Failed to get metrics from http://127.0.0.1:8080/api/v3/nodes/[email protected]/metrics/ source="emq_exporter.go:98"
WARN[0007] Failed to get metrics from http://127.0.0.1:8080/api/v3/nodes/[email protected]/metrics/ source="emq_exporter.go:98"
WARN[0009] Failed to get metrics from http://127.0.0.1:8080/api/v3/nodes/[email protected] source="emq_exporter.go:98"

Type prefix

I found that emq-exporter's own data type has no prefix. It is recommended to add a uniform type prefix to prevent confusion. Uniform prefixes can also be retrieved in the Grafana Dashboard.

emq_nodes_stats_topics_max 6
# HELP emq_up Was the last scrape of EMQ successful
# TYPE emq_up gauge
emq_up 1
# HELP go_gc_duration_seconds A summary of the GC invocation durations.
# TYPE go_gc_duration_seconds summary
go_gc_duration_seconds{quantile="0"} 0.000201694
go_gc_duration_seconds{quantile="0.25"} 0.000232821
go_gc_duration_seconds{quantile="0.5"} 0.00026052
go_gc_duration_seconds{quantile="0.75"} 0.000269629

multiple nodes support

Is there an idea to support multiple nodes? The current version only supports a single node, but the emqx official v3 version of the api can get the node list.

Authorization issue

I had to add this exporter not as a side-car, but as a separate deployment(unfortunately this was a requirement).

We've deployed EMQX as a "EMQX Kubernetes Operator"

My deployment configuration:

image: nuvo/emq_exporter:v0.5.1
  - args:
     - -emq.uri=http://emqx-0.emqx-svc-headless.core.svc.cluster.local:8081
     - [email protected]
     - -emq.api-version=v4

• credentials are injected as env vars through the secret, that is automatically created by kubernetes operator

ISSUE: The exporter container log says:

{"level":"warn","time":1670324415,"caller":"/home/travis/gopath/src/github.com/nuvo/emq_exporter/emq_exporter.go:76","message":"Received status code not ok http://emqx-0.emqx-svc-headless.core.svc.cluster.local:8081/api/v4/nodes/[email protected]/metrics/, got 401"}

"emq_nodes_metrics_packets.publish.received" is not a valid metric name

Hello, I am getting these kind of errors since updating to emqx v3.2.0. From what I can see this is just a matter of emqx publishing metric names with . instead of / , should be an easy fix. Let me know if you are still maintaining this, I can do a pull request.

Thank you,
Radu

emq exporter api v4

Hi,

I am testing emq exporter (prometheus) for monitoring. my emq has api version 4. this exporter doesnt support v4. is there any work around ?

Have a template of grafana dashboard for monitoring emq？

I am using grafana + prometheus monitoring emq，but there is no avaiable template, any suggest?

topic name

I recommend adding the topic name to the metrics. Only the number of topics is too small to get the information. You can use the topic name as the label to classify the data.

add prometheus label.

I recommend using the label function of prometheus for similar data aggregation. Label is very useful for query and data calculations. The current data items are independent, and configuring the dashboard is not convenient.