We suffered recently from a network switch issue in a chassis used for CI infra.
As that chassis is out of warranty, there was no way to have support nor the network switch replaced.
The workaround to restore service was to consolidate some network bridges on all nodes to be running from eth1 (as eth0 had no connectivity anymore)

But that's blocking currently redeploying/reinstalling some nodes through ansible automation, as supermicro nodes can only do pxe/network boot from eth0.

The plan:

• reconfigure (but not reload) network config/bridges on hosts to go back to eth0
• shutdown all nodes
• work with PNT/Devops team to swap network switches in chassis (so that eth0 is connected and eth1 is on bad switch)
• validate
• restart all nodes