Alertmanager CRD overlaps with the Prometheus Operator one

[clmssz]

Hi there,

Not sure if the files in master are the most up to date ones, since you mention an install directory in the docs and it's not there yet, but in the deploy/crds for vmalertmanagers, singular and plural terms mismatch - If the goal is to use the coreos CRD for alertmanager, better remove it from CRDs here as it will conflict with an already deployed prometheus operator

  names:
    kind: Alertmanager
    listKind: AlertmanagerList
    plural: vmalertmanagers
    shortNames:
    - vma
    singular: alertmanager

[f41gh7]

Hello, thx for report.

Its the bug, at an early version alertmanager was at coreos api, later ive moved it into victoriametrics api.

I think it`s better to rename it to VmAlertmanager and fix the conflict.

[clmssz]

Candid question : Is there a reason not to use the Coreos one ? Like a victoria alertmanager in the making ? Or you want to separate from Coreos APIs ?
Asking because for us, the use of PrometheusRule, ServiceMonitors etc.. as "standard" is really practical, since we will be migrating from prometheus operator deployments.

[f41gh7]

The main reason is the conflict with existing prometheus-operator setup. If you run both operators at one cluster it would result in race-condition updates for alertmanager (at coreos api) and may lead to unwanted behaviour.
Plus configuration at victoriametrics api for alertmanager is a bit differ - extraEnv,extraArgs,replicaCount.

[clmssz]

Got it

[f41gh7]

Renamed to VMAlertmanager, now it's :

  group: victoriametrics.com
  names:
    kind: VMAlertmanager
    listKind: VMAlertmanagerList
    plural: vmalertmanagers
    shortNames:
    - vma
    singular: vmalertmanager