[Bug] prometheus metric format is invalid

[qwtsc]

Bug Type (问题类型)

None

Before submit

• 我已经确认现有的 Issues 与 FAQ 中没有相同 / 重复问题 (I have confirmed and searched that there are no similar problems in the historical issue and documents)

Environment (环境信息)

hugegraph in k8s using prometheus and grafana as service monitor.

Expected & Actual behavior (期望与实际表现)

prometheus metrics format is invalid. After some investigations and code review, I think that the problem would probably be that this PR do not use the correct data exposition format that prometheus community required. The acceptable histogram format can be viewed as follows:

# A histogram, which has a pretty complex representation in the text format:
# HELP http_request_duration_seconds A histogram of the request duration.
# TYPE http_request_duration_seconds histogram
http_request_duration_seconds_bucket{le="0.05"} 24054
http_request_duration_seconds_bucket{le="0.1"} 33444
http_request_duration_seconds_bucket{le="0.2"} 100392
http_request_duration_seconds_bucket{le="0.5"} 129389
http_request_duration_seconds_bucket{le="1"} 133988
http_request_duration_seconds_bucket{le="+Inf"} 144320
http_request_duration_seconds_sum 53423
http_request_duration_seconds_count 144320

and I do also think the correct metric type here is supposed to be summary rather than histogram according to the prometheus best practice documentation.

By the way, I am willing to provide a PR to fix this issue as well as grafana dashboard configuration after I figure it out .

Vertex/Edge example (问题点 / 边数据举例)

No response

Schema [VertexLabel, EdgeLabel, IndexLabel] (元数据结构)

No response

[qwtsc]

And metric name is also invalid. There are not allowed to have / or $, so this snippet of code should change to the following.

    public static String replaceDotDashInKey(String orgKey) {
        return orgKey.replace(".", "_").replace("-", "_").replace("/", "_").replace("$", "_");
    }

[JackyYangPassion]

Hi @qwtsc , thanks for reporting the issue.
Welcome to submit PR for fixes

[imbajin]

@SunnyBoy-WYH could u check the issue when free

[SunnyBoy-WYH]

@qwtsc Thanks for your report, welcome to submit PR ; but can you introduce the request path /param and something else that we can recurrent it ?

ref: https://hugegraph.apache.org/docs/clients/restful-api/metrics/#1-basic-metrics

[qwtsc]

@SunnyBoy-WYH the request path is just /metrics. check the following prometheus service monitor.

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: prometheus-hugegraph
  # Change this to the namespace the Prometheus instance is running in
  namespace: default
  labels:
    app: hugegraph
    release: prometheus
spec:
  selector:
    matchLabels:
      name: hugegraph
  namespaceSelector:
    matchNames:
    - hugegraph-e2e
  endpoints:
  - port: metrics
    path: /metrics
    interval: 15s

[SunnyBoy-WYH]

@SunnyBoy-WYH the request path is just /metrics. check the following prometheus service monitor.

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: prometheus-hugegraph
  # Change this to the namespace the Prometheus instance is running in
  namespace: default
  labels:
    app: hugegraph
    release: prometheus
spec:
  selector:
    matchLabels:
      name: hugegraph
  namespaceSelector:
    matchNames:
    - hugegraph-e2e
  endpoints:
  - port: metrics
    path: /metrics
    interval: 15s

hi, will you continue to fix this issue ? @qwtsc

[SunnyBoy-WYH]

how about this？ but now we just let admin read metric, so we can config auth in prometheus.yml.; anything question?

[SunnyBoy-WYH]

how about this？ but if we open hg auth mode, we only let admin read metric, so we can config auth in prometheus.yml.; anything question?