postgres-operator pod crashed when delete instance which enable or disable connect pooler

[godzilla-s]

Please, answer some short questions which should help us to understand your problem / question better?

• Which image of the operator are you using? e.g. registry.opensource.zalan.do/acid/postgres-operator:v1.7.1
• Where do you run it - cloud, k3s
• Are you running Postgres Operator in production? [yes]
• Type of issue? [Bug report]

Hello:
I found uncertain issue: the operator pod crashed when delete instance that enable and disable connect pooler, the step as below:

1. create a instance, enable connect pooler

enableConnectionPooler: true
enableReplicaConnectionPooler: true

it works fine after it created

2. then update instance: disable connect pooler

enableConnectionPooler: false
enableReplicaConnectionPooler: false

it works fine after updated

3. delete the instance, a few seconds later, the postgres-operator pod crashed, the logs like:

panic: rutnime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x1a76754]

github.com/zalando/postgres-operator/pkg/cluster.(*Cluster).deleteSecrets
          /workspace/pkg/cluster/resources.go: 502 +0x154
github.com/zalando/postgres-operator/pkg/cluster.(*Cluster).Delete
         /workspace/pkg/cluster/cluster.go:903 +0x256

and I found some problem with the code:

func (c *Cluster) deleteSecrets() error {
	c.setProcessName("deleting secrets")
	var errors []string
	errorCount := 0
	for uid, secret := range c.Secrets {
               // check if secret is nil ???????
                // if not verify, it will crashed  ????
		err := c.deleteSecret(uid, *secret)
		if err != nil {
			errors = append(errors, fmt.Sprintf("%v", err))
			errorCount++
		}
	}

	if errorCount > 0 {
		return fmt.Errorf("could not delete all secrets: %v", errors)
	}

	return nil
}

func (c *Cluster) deleteSecret(uid types.UID, secret v1.Secret) error {
	c.setProcessName("deleting secret")
	secretName := util.NameFromMeta(secret.ObjectMeta)
	c.logger.Debugf("deleting secret %q", secretName)
	err := c.KubeClient.Secrets(secret.Namespace).Delete(context.TODO(), secret.Name, c.deleteOptions)
	if err != nil {
		return fmt.Errorf("could not delete secret %q: %v", secretName, err)
	}
	c.logger.Infof("secret %q has been deleted", secretName)
	c.Secrets[uid] = nil  // set to nil?  or just delete(c.Secrets, uid) ?

	return nil
}

I hope some one can help to make sure the issue?

[FxKu]

Thanks @godzilla-s for raising this. I could reproduce this bug. Can you provide a PR using delete(c.Secrets, uid) instead of c.Secrets[uid] = nil?

[godzilla-s]

hello, thanks for your reply, I have test it and submit PR , please verify it @FxKu

[FxKu]

fixed in 1.9.0