metrics: change the port for metrics-exporter

[weirdwiz]

in host mode, metrics exporter has a bind contention with kubevirt so we
need to change the port to something other than 8080 / 8081

fixes: DFBUGS-5812

[weirdwiz]

ptal @yati1998 @aruniiird @malayparida2000

[yati1998]

any specfic reason for changing the position, I dont see any relevant change here

[aruniiird]

Creating the Service first ensures the endpoints are ready when pods start.

[weirdwiz]

yes, just extra protection for race conditions

[SanjalKatiyar]

@weirdwiz will this change always prevent such port conflicts ??

Example (in case of host networking): say if two different deployments (both using kube-rbac-proxy) are scheduled on the same node, wouldn't they conflict if they expose the same proxy ports (8443 or 9443). Will we then randomly change these ports to something else too ??
Similarly, in this PR, even with the exporter ports changed to new values, other components could still choose to use those ports (just less likely, but still possible).

So, isn't this a wider issue ?? How do other ODF components handle such cases ?? Or is there something I might be missing here ??

[weirdwiz]

@weirdwiz will this change always prevent such port conflicts ??

Example (in case of host networking): say if two different deployments (both using kube-rbac-proxy) are scheduled on the same node, wouldn't they conflict if they expose the same proxy ports (8443 or 9443). Will we then randomly change these ports to something else too ?? Similarly, in this PR, even with the exporter ports changed to new values, other components could still choose to use those ports (just less likely, but still possible).

So, isn't this a wider issue ?? How do other ODF components handle such cases ?? Or is there something I might be missing here ??

this PR isn't trying to solve arbitrary port conflicts. it's fixing a specific, reported conflict (DFBUGS-5812) where kubevirt binds 8080/8081 on the host, and in host networking mode the exporter shares the same network namespace so they collide. other ODF components handle this the same way for example, RGW switches from port 80/443 to 50080/50443 when host networking is enabled

[weirdwiz]

ptal @Madhu-1

[SanjalKatiyar]

this PR isn't trying to solve arbitrary port conflicts. it's fixing a specific, reported conflict (DFBUGS-5812) where kubevirt binds 8080/8081 on the host, and in host networking mode the exporter shares the same network namespace so they collide.

I understand that.
What I was asking, if we encounter a similar issue again (like examples in my previous comment), we would again pick a different/random port and hope it isn't already used on the host, that's somewhat arbitrary.

other ODF components handle this the same way for example, RGW switches from port 80/443 to 50080/50443 when host networking is enabled

That's what I meant by wider issue in my previous comment. But if this is already what we do across ODF components, and it's something we have already discussed and agreed on, then that answers my question.

[weirdwiz]

/cherry-pick release-4.21

[openshift-cherrypick-robot]

@weirdwiz: once the present PR merges, I will cherry-pick it on top of release-4.21 in a new PR and assign it to you.

Details

In response to this:

/cherry-pick release-4.21

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: malayparida2000, weirdwiz

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [malayparida2000]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[weirdwiz]

/cherry-pick release-4.21