Export currentOP uptime query metrics

[tregubov-av]

Resolves: #704

This is the implementation of the collector for notifications about too long requests are in processed.
For example:

apiVersion: v1
data:
  kube-state-metrics-mongodb.rules: |-
    groups:
    - name: kube-state-metrics-mongodb.rules
      rules:
      - alert: MongodbCurrentQueryTime
        expr: (mongodb_currentop_query_uptime > 3e+8) / 1000
        labels:
          severity: critical
        annotations:
          description: "Opid: {{ $labels.opid }}\nDesc: {{ $labels.desc }}\nNs: {{ $labels.ns }}\nOp : {{ $labels.op }}\nUptime : {{ $value }} ms\n"
          summary: "MongoDB\nCurrent slow query on: {{ $labels.endpoint }}"
kind: ConfigMap
metadata:
  labels:
    app: prometheus
    prometheus: kube-prometheus
    release: kube-prometheus-stack
    role: alert-rules
  name: kube-prometheus-exporter-mongodb
  namespace: monitoring

---

• Tests passed.
• Fix conflicts with target branch.

[it-percona-cla]

All committers have signed the CLA.

[BupycHuk]

Thanks @tregubov-av,

Looks good to me.
Could you please sign the CLA?

[BupycHuk]

do we really need all these assertions?

[tregubov-av]

I think for a more flexible construction of alerts, you need to have a sufficient number of labels.
In this case, you have to use type assertion to get the values from the interface.
But I don't want to panic if the code fails.
If there are suggestions on how to do better, I'm ready to rewrite this part

[tregubov-av]

This is weird...
Locally all tests pass on all available versions MongoDB (4.2, 4.4, 5.0)
What am I doing wrong?
Prepared a pull request based on this documentation https://github.com/percona/mongodb_exporter/blob/main/docs/development-guide.md

P.S.
I think I understand what the problem is with passing integration tests.
Perhaps they are launched immediately after the docker-compose has completed, but the MongoDB cluster has not yet had time to start.
Locally, this behavior is reproduced if immediately after make test-cluster runs make test.
I came to this idea because the tests of those controllers to which I did not make changes ended with errors.

[BupycHuk]

Hi @tregubov-av, yeah, problem isn't in your PR. We started having this problem for all PRs recently and thanks for inputs. we will try to fix that.

[tregubov-av]

👍