[Bug]: Operator deletes PodMonitor that it didn't create due to lack of owner reference check

Is there an existing issue already for this bug?

I have searched for an existing issue, and could not find anything. I believe this is a new bug.

I have read the troubleshooting guide

I have read the troubleshooting guide and I think this is a new bug.

I am running a supported version of CloudNativePG

I have read the troubleshooting guide and I think this is a new bug.

Contact Details

No response

Version

1.24.0

What version of Kubernetes are you using?

1.30

What is your Kubernetes environment?

Cloud: Google GKE

How did you install the operator?

YAML manifest

What happened?

[Bug]: Prometheus v2.52+ fails to scrape metrics: Error on ingesting samples with different value but same timestamp #6108 , decide to set monitoring.enablePodMonitor: false. In the same commit, create a PodMonitor to be applied manually, with the same name (i.e. the cluster name) and namespace.
Apply both changes simultaneously
Wonder why I don't have a PodMonitor
Go spelunking in the code:

cloudnative-pg/internal/controller/cluster_create.go

Line 959 in 3c2c3f6

client.ObjectKeyFromObject(expectedPodMonitor),

gets a reference to the PodMonitor based on namespace and name only, not looking at the owner reference. Therefore

cloudnative-pg/internal/controller/cluster_create.go

Line 973 in 3c2c3f6

case !manager.IsPodMonitorEnabled() && podMonitor != nil:

indeed finds the PodMonitor (that it doesn't have an owner reference on) and deletes it.

I worked around the issue by adding a short suffix to the PodMonitor name so that it wouldn't be deleted, but this isn't ideal because it changes the job label that Prometheus attaches to it.

Cluster resource

N/A

Relevant log output

N/A

Code of Conduct

I agree to follow this project's Code of Conduct

I'm also impacted by this issue.

Note that if you need to create your own PodMonitor but want to keep the same job label, you can override it using relabelings:

apiVersion: monitoring.coreos.com/v1
kind: PodMonitor
metadata:
  name: clustername-suffix
spec:
  podMetricsEndpoints:
  - relabelings:
    - replacement: clustername
      targetLabel: job
    port: metrics
  selector:
    matchLabels:
      cnpg.io/cluster: clustername

This issue is stale because it has been open for 60 days with no activity.

bump to keep open

Have the same problem

As project maintainers, we have decided to deprecate the native integration with PodMonitor resources starting from version 1.28. We will delegate all functionality to the official upstream PodMonitor implementation. For more details, please take a look at issue #8075.

I get that you want to deprecate it but right now it is deleting my custom resources even if the operator did not create them in the first place.

This is happening for me too

Issue is still happening, hopefully #9340 gets merged soon.

fix: podmonitor name cant match cluster name

we should probably update the cnpg docs so they don't have explicit examples showing configs where the podmonitors will get deleted. will try to put together a PR. personally dont think we need the relabeling for doc examples, i didnt think the CNPG official dashboard had any references to it

fix: podmonitor name cant match cluster name

docs: Update PodMonitor examples to use distinct names from clusters

fix: add ownership check before deleting PodMonitor (#9340)

fix: podmonitor name cant match cluster name (#57)

fix: add ownership check before deleting PodMonitor (#9340)

Navigation Menu

Repository navigation

[Bug]: Operator deletes PodMonitor that it didn't create due to lack of owner reference check #6109

Is there an existing issue already for this bug?

I have read the troubleshooting guide

I am running a supported version of CloudNativePG

Contact Details

Version

What version of Kubernetes are you using?

What is your Kubernetes environment?

How did you install the operator?

What happened?

Cluster resource

Relevant log output

Code of Conduct

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

[Bug]: Operator deletes PodMonitor that it didn't create due to lack of owner reference check #6109

Description

Is there an existing issue already for this bug?

I have read the troubleshooting guide

I am running a supported version of CloudNativePG

Contact Details

Version

What version of Kubernetes are you using?

What is your Kubernetes environment?

How did you install the operator?

What happened?

Cluster resource

Relevant log output

Code of Conduct

Activity

Jean-Daniel commented on Feb 27, 2025

github-actions commented on May 18, 2025

sorrison commented on May 18, 2025

SeaRoll commented on Jul 8, 2025

gbartolini commented on Jul 21, 2025

runningman84 commented on Nov 22, 2025

veshant commented on Nov 24, 2025

christiaangoossens commented on Dec 21, 2025

ardentperf commented on Jan 15, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions