octavia/ovn: missed healthmon port cleanup

Test script to reproduce the issue.
Can be run against any OpenStack environment with octavia ovn provider loadbalancers without any special privileges.
But was originally observed against OpenStack 2023.2 (Bobcat) as configured by OSISM which uses kolla-ansible.
Original report is here:
https://github.com/osism/issues/issues/921

Here's a patch against the ovn-octavia helper that addressed this.
What it does:
* Whenever a port is deleted and a health-mon is active, it does check whether this was the last port in this subnet.
* If so, the neutron hm port does get deleted.

This fixes the test case.

moving to neutron, the ovn-octavia-provider is a neutron project

The wrong condition checking here on member_delete [1] about pool == OFFLINE is the condition breaking the logic, because other members from diff subnet could keep the pool != OFFLINE and if no more member are there from subnetX, the hm_port for subnetX will be a leftover.

The patch provider in c#2 is wrong, it looks fixing the test case because is calling to _clean_up_hm_port everytime a member is deleted, and _clean_up_hm_port is taking care of delete or not according to the pending members.

I will provide a patch (mostly similar to your proposal) in a few moments to fix this corner case.

Kurt, for next time, send your patch upstream and we'll work together on that patch, it would be a pleasure to have more people contributing to the project ;)

[1] https://opendev.org/openstack/ovn-octavia-provider/src/branch/master/ovn_octavia_provider/helper.py#L2200

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/ovn-octavia-provider/+/916637

Thanks, Fernando, I will carefully read your patch to spot the difference.
I would really appreciate to see the fix also in caracal and bobcat to make octavia/ovn useful there.
When you say "send your patch upstream next time": You prefer mails to openstack-dev rather than bug reports with reproduction scripts and patches?

You're welcome Kurt, once merged into master, we can backport to stable branches. Regarding my comment, I meant that you can propose your fixs/improvements to the code just like I sent the patch upstream yesterday, basically with a new patch to the repository via https://review.opendev.org/.

And of course, reporting bugs here on Launchpad, and if they come with a reproducer script, it's also very helpful.

Reviewed: https://review.opendev.org/c/openstack/ovn-octavia-provider/+/916637
Committed: https://opendev.org/openstack/ovn-octavia-provider/commit/f034bab144b68cf96c538339e389c4cc7c6d7d63
Submitter: "Zuul (22348)"
Branch: master

commit f034bab144b68cf96c538339e389c4cc7c6d7d63
Author: Fernando Royo <email address hidden>
Date: Mon Apr 22 15:47:46 2024 +0200

    Remove leftover OVN LB HM port upon deletion of a member

    When a load balancer pool has a Health Monitor associated with it,
    an OVN LB Health Monitor port is created for each backend member
    subnet added.

    When removing backend members, the OVN LB Health Monitor port is
    cleaned up only if no more members are associated with the Health
    Monitor pool. However, this assumption is incorrect. This patch
    corrects this behavior by checking instead if there are more members
    from the same subnet associated with the pool. It ensures that the
    OVN LB Health Monitor port is deleted only when the last member from
    the subnet is deleted. If the port is being used by another different
    LB Health Monitor, `_clean_up_hm_port` will handle it.

    Closes-Bug: #2062965
    Change-Id: I4c35cc5c6af14bb208f4313bb86e3519df0a30fa

Fix proposed to branch: stable/2024.1
Review: https://review.opendev.org/c/openstack/ovn-octavia-provider/+/917724

Fix proposed to branch: stable/2023.2
Review: https://review.opendev.org/c/openstack/ovn-octavia-provider/+/917725

Fix proposed to branch: stable/2023.1
Review: https://review.opendev.org/c/openstack/ovn-octavia-provider/+/917726