Description of problem:
After the upgrade of 7 RHOSP envs from 13 to 16.2 during 2024, cu found out that the upgrade changed the router distribution on the Controllers, on all RHOSPs:

• Before upgrade: routers were fairly distributed between the 3 Controllers
• After upgrade: almost all routers are placed on 1 Controller (the boostrap one).

From metrics the Controller hosting most of the routers has much more network traffic than the 2 other Controllers, so that can limit network bandwidth.

This issue concerns only the old routers (created on RHOSP13). New tenants with new routers (on RHOSP16) are faily distributed.

Current situation on one env from ovn-nbctl lrp-get-gateway-chassis lrp-<id> perspective the number Chassis with priority 3 are:
chassis-1 10 entries
chassis-2 10 entries
chassis-3 96 entries

Version-Release number of selected component (if applicable):
[rhosp-release] Red Hat OpenStack Platform release 16.2.6 (Train

How reproducible:
FFU from 13 to 16.2

Actual results:
The router went all on same router

Expected results:
The router should be distributed controller nodes.

Additional info:
The customer asks for a workaround for change the priorities, in order to move rebalance the situation