Signed-off-by: Zenghui Shi <zshi@redhat.com>

In the DPU 2-clusters design, a host with DPU will join the cluster
with ovn-controller running in the host first, then it will be moved
to the DPU. Therefore, we will have 2 chassis with the same hostname
in the ovn sbdb.

With this patch, the master will try to remove the stale chassis when
the Chassis ID is inconsistent with the value of node annotation
'k8s.ovn.org/node-chassis-id'.

Signed-off-by: Peng Liu <pliu@redhat.com>

Bump libovsdb to include ovn-org/libovsdb#321

Signed-off-by: Numan Siddique <numans@ovn.org>

Currently, they are reported in milliseconds.
Prometheus mandates time values are reported in seconds.

Signed-off-by: Martin Kennelly <mkennell@redhat.com>

OVN-K Metrics: Ensure stopwatch metrics are reported in seconds

Report OVN controller connection status via a metric from each
OVN kube node.

The metric is updated every 2 minutes in-order not to incur
polling overhead which is not insignificant.

Prometheus scrape interval is usually every 30 seconds, and is
this value on OCP.

The reason we report this from every node instead of grabbing it
from SB DB via (nb_cfg trigger), is to save SB DB from each
OVN controller having to write to it.
This maybe an issue when scaling node.

Signed-off-by: Martin Kennelly <mkennell@redhat.com>

Signed-off-by: Martin Kennelly <mkennell@redhat.com>

Add metrics to observe core events related to EgressIP:

- name: egress_ips_assign_latency_seconds -- Histogram **
  desc: The latency of egress IP assignment to ovn nb database

- name: egress_ips_unassign_latency_seconds -- Histogram **
  desc: The latency of egress IP unassignment from ovn nb database

- name: egress_ips_node_unreachable_total -- Counter
  desc: The total number of times assigned egress IP(s) were unreachable

- name: egress_ips_rebalance_total -- Counter
  desc: The total number of times assigned egress IP(s) needed to be moved to a different node

Add flags to explicitly enable the histogram metrics, since we only see value
in having them when scale testing egress ips. The flag introduced here is:
    --metrics-enable-eip-scale

Signed-off-by: Flavio Fernandes <flaviof@redhat.com>

This reverts commit 503de2f.

Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>

…ube-node"

This reverts commit 1a225ea.

Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>

…metric

OVN-K metrics: Add OVN controller southbound database connection

Bumps [@actions/core](https://github.com/actions/toolkit/tree/HEAD/packages/core) from 1.2.6 to 1.9.1.
- [Release notes](https://github.com/actions/toolkit/releases)
- [Changelog](https://github.com/actions/toolkit/blob/main/packages/core/RELEASES.md)
- [Commits](https://github.com/actions/toolkit/commits/HEAD/packages/core)

---
updated-dependencies:
- dependency-name: "@actions/core"
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>

Remove stale chassis for hosts that run ovnkube-node on DPU

Revert EndpointSlice commits

This deletes any stale ip from node allocations map before assignment
so that update on same egress ip object would assign new ip address
on a particular assignable node in any timing scenarios.

Signed-off-by: Periyasamy Palanisamy <pepalani@redhat.com>

Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>

fedora Dockerfile : Switch to f36 and OVN 22.06.

…hub/actions/cleanup-action/actions/core-1.9.1

Bump @actions/core from 1.2.6 to 1.9.1 in /.github/actions/cleanup-action

egressip: add metrics

libovsdb not being able to reset ACl severity to nil), use nil to reset.
Unify ACL logging setup (ovn/acl.go) to make ACL severity properly
set and reset (since severity is *string, can be error-prone)
Add unified handling for ACL logging that is always disabled (
like arpAllowACL)

Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>

Add egress firewall external ID constant.

Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>

namespace update handler will only update them if
len(nsInfo.networkPolicies) > 0, which is not the case during the first
ns policy creation - it can lead to default ACLs loglevel not being
updated.

Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>

Signed-off-by: Pardhakeswar Pacha <ppacha@nvidia.com>
(cherry picked from commit 1a225ea)

Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>
(cherry picked from commit 503de2f)

The selector in the endpointslice informer now correctly chooses only endpointslices with a non-empty "kubernetes.io/service-name" label and without "service.kubernetes.io/headless" label.

Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>

Remove default ACL severity

OCPBUGS-417: Fix informer selector for endpointslices

When the logical port is already deleted, deleteLogicalPort should
not return an error. Otherwise, delete will be retried up for
maxFailedAttempts (currently set to 15).

This issue introduced in commit be8786a

Signed-off-by: Flavio Fernandes <flaviof@redhat.com>
Co-authored-by: Tim Rozet <trozet@redhat.com>
Reported-at: https://issues.redhat.com/browse/OCPBUGS-469

pods: deleteLogicalPort should not fail when port is already gone

Delete stale egress ip before assigning new ip

…for ovnkube-node""

This reverts commit 69942d0.

This reverts commit bb41f5e.

…eam_merge