Stop using legacy network interface names (eth0)

[bdunne]

It's no longer working for GCE and Azure appliances, they mark the connection as unmanaged and won't allow it to come up. Also, our scripts no longer depend on the interface name in EL9.

net.ifnames=0 reverts to legacy adapter names (eth0)
biosdevname does the same for Dell systems and requires a package we don't have installed
network-scripts are no longer used by NetworkManager

[miq-bot]

Checked commit bdunne@f9c43db with ruby 3.1.5, rubocop 1.56.3, haml-lint 0.51.0, and yamllint
0 files checked, 0 offenses detected
Everything looks fine. 🍪

[Fryguy]

Thanks for figuring this out @bdunne

cc @agrare thought you might be interested.

[Fryguy]

LGTM, but would like a second set of eyes. @agrare ?

[bdunne]

I tested removing these kernel args from both Azure and GCP appliances locally and the interfaces came up.

[bdunne]

May also be related to ManageIQ/manageiq#22973

[agrare]

Love it, and great that "doing the right thing" fixes these issues also. We do still assume eth0 in a few places that we'll need to update:

manageiq/app/models/miq_server/environment_management.rb:          eth0 = LinuxAdmin::NetworkInterface.new("eth0")
manageiq-appliance_console/bin/appliance_console:  eth0 = LinuxAdmin::NetworkInterface.new(NETWORK_INTERFACE)

[agrare]

There is one more place in appliance_console with eth0, database replication:

lib/manageiq/appliance_console/database_replication_primary.rb:      self.primary_host      = LinuxAdmin::NetworkInterface.new(NETWORK_INTERFACE).address
lib/manageiq/appliance_console/database_replication_standby.rb:      self.standby_host      = LinuxAdmin::NetworkInterface.new(NETWORK_INTERFACE).address
lib/manageiq/appliance_console/database_replication.rb:    NETWORK_INTERFACE = 'eth0'.freeze

Okay I've updated ManageIQ/manageiq-appliance_console#263 to take care of the database replication issue.

[bdunne]

Ok, I think all related PRs are merged now.