Bug 2219405

Summary: Disable LSP_OPTIONS_MCAST_FLOOD_REPORTS for LSPs that may negatively affect the length of action lists in mcast handling flows
Product: Red Hat OpenStack Reporter: Ihar Hrachyshka <ihrachys>
Component: python-networking-ovnAssignee: Lucas Alvares Gomes <lmartins>
Status: CLOSED CURRENTRELEASE QA Contact: Eran Kuris <ekuris>
Severity: high Docs Contact:
Priority: high    
Version: 16.2 (Train)CC: apevec, ftaylor, lhh, majopela, ralonsoh, scohen
Target Milestone: ---Keywords: Triaged
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2024-09-05 08:22:13 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Ihar Hrachyshka 2023-07-03 13:51:04 UTC 
This bug was initially created as a copy of Bug #2218631

I am copying this bug because: 

We should follow up on the comment from the original bug: https://bugzilla.redhat.com/show_bug.cgi?id=2218631#c18

Specifically, we should disable LSP_OPTIONS_MCAST_FLOOD_REPORTS for ports created by ml2/ovn: https://github.com/openstack/neutron/blob/b92d133de66411b9444855cc97480891d7d7b7b2/neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovn_client.py#L369

The hope is that this will alleviate the size of action lists in table=38 of OVN (for the flows that handle mcast flooding), making it a potential work around for the OVS action list size limitation (tracked in the original bug).

Description of problem:
After creating close to 1220 VMs on a single compute, networking breaks for all of them and we see the following errors in the logs:

Jun 29 17:20:57 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:20:57 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:20:59 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:20:59 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:00 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:07 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:07 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:07 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:07 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:08 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:08 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:08 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:13 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:13 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:14 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:14 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:15 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768
Jun 29 17:21:15 compute00 kernel: openvswitch: netlink: Flow action size exceeds max 32768


Version-Release number of selected component (if applicable):
RHOSP16.2.5

How reproducible:
Always

Steps to Reproduce:
1. Create close to 1220 VMs (at 1219 it should still work)
2.
3.

Actual results:
Global outage

Expected results:
Create 2000 vMs on a single compute

Additional info: