Wiki -> https://fedoraproject.org/wiki/Changes/EnableConsistentDeviceNamingCloud
This document represents a proposed Change. As part of the Changes process, proposals are publicly announced in order to receive community feedback. This proposal will only be implemented if approved by the Fedora Engineering Steering Committee. = Enable Consistent Device Naming in Cloud Images = == Summary == This proposal aims to remove the `net.ifnames=0` kernel command line entry from the Fedora cloud kickstarts so that consistent device naming is enabled for cloud instances. This change brings Fedora Cloud in line with Fedora Server, Workstation, and CoreOS. == Owner == * Name: [[User:mhayden|Major Hayden]] * Email: ma...@redhat.com == Detailed Description == Fedora cloud images currently set `net.ifnames=0` on the kernel command line during the [https://pagure.io/fedora-kickstarts/blob/main/f/fedora-cloud-base.ks#_35 kickstart] process. This disables consistent device naming and ensures that ethernet devices retain the old-style names of `eth0`, `eth1`, `eth2`, and so on. Removing the `net.ifnames=0` configuration allows Fedora cloud instances to use consistent device names for network devices. This brings Cloud images in line with Fedora Server, Workstation, and CoreOS. == Feedback == One of the proposed alternatives is to leave `net.ifnames=0` in the kernel command line to remain consistent between releases. Although RHEL allows for `net.ifnames=0` for KVM instances, it is [https://access.redhat.com/solutions/2435891|strongly recommended not to use it] with OpenStack or RHV environments. Fedora Cloud images are used for multiple types of clouds, including public clouds and private clouds. This approach is less disruptive, but it pushes off the consistent device naming change until a later date and causes cloud images to operate differently than other Fedora deployments. Other distributions have taken different turns with this configuration. [https://lists.ubuntu.com/archives/ubuntu-devel/2016-April/039325.html| Ubuntu] removed it in 2016, Debian [https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=863385| added it back in 2017], and [https://documentation.suse.com/sles/15-SP4/pdf/article-minimal-vm_en.pdf| SUSE has had it] for some time. There were issues with cloud-init during the early days of network device naming in 2016/2017, but [https://github.com/canonical/cloud-init/blob/d8e3a4b4d4ca443227bf06bdaa5d9b47bd84e326/cloudinit/net/__init__.py#L480-L491| cloud-init supports consistent network names] since 2017. RHEL currently uses `net.ifnames=0`, but there is a debate about whether it should be removed in a future major release. == Benefit to Fedora == The main benefit comes from bringing cloud instances inline with other deployment types in how they name network devices. It also helps with cloud providers that offer up different types of network interfaces to an instance. For example, an emulated network device would present itself differently than one offered via SRIOV and it would allow an instance administrator to easily tell which network interface corresponds to each network device. This is especially important for certain clouds, such as OpenStack clouds. == Scope == * Proposal owners: * Other developers: * Release engineering: [https://pagure.io/releng/issues #Releng issue number] * Policies and guidelines: N/A (not needed for this Change) * Trademark approval: N/A (not needed for this Change) * Alignment with Community Initiatives: == Upgrade/compatibility impact == '''Upgrades:''' Users who are upgrading to the next Fedora release will not notice a change in their instances since the `net.ifnames=0` change is only applied during the kickstart process. Their instances will continue using the old network names. '''New deployments:''' If a user has older Fedora deployments and they deploy a new Fedora release with this change applied, their network devices will use consistent network names instead of the old `eth0` and `eth1` style names. Although this won't impact software like cloud-init, it will impact users who have deployment scripts (Terraform or Ansible, for example) that need to set network configuration based on the network adapter's name. They will need to adjust the name of the network device in their deployment scripts. == How To Test == Use an image built without `net.ifnames=0` or take an existing cloud instance and remove `net.ifnames=0` from the kernel command line: grubby --update-kernel=current_kernel --remove-args="kernel_args" After a reboot, the device names should change from `eth0` to reflect the consistent device names based on the hardware: $ ip link | grep ^[0-9] 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000 2: enp2s0f0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN mode DEFAULT group default qlen 1000 3: eno1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP mode DEFAULT group default qlen 1000 4: enp2s0f1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN mode DEFAULT group default qlen 1000 == User Experience == Users who don't need to apply any specialized network configuration will likely not notice a change. All of the cloud-init functionality will work the same as it did previously. For users who need to apply specific network configuration on specific instances will need to change the name they use to refer to those network interfaces. For example, the main network device may shift from `eth0` to `ens3` on some KVM platforms. Other cloud instances might show a network device that specifies a PCI bus entry, such as `enp2s0f0`. Although this might be confusing at first glance, it will be familiar to users of Fedora Workstation, Server, and CoreOS. Users could opt out by adding `net.ifnames=0` back to their kernel command line: grubby --update-kernel=ALL --args="net.ifnames=0" == Dependencies == This change does not depend on any other changes and it has no impact on cloud-specific dependencies, such as cloud-init. If users have custom setup scripts that apply network configuration, or if they use Terraform/Ansible to perform these steps, they may need to adjust their scripts to account for the new network names. == Contingency Plan == * Contingency mechanism: (What to do? Who will do it?) N/A (not a System Wide Change) * Contingency deadline: N/A (not a System Wide Change) * Blocks release? N/A (not a System Wide Change), Yes/No == Documentation == N/A (not a System Wide Change) == Release Notes == -- Aoife Moloney Fedora Operations Architect Fedora Project Matrix: @amoloney:fedora.im IRC: amoloney -- _______________________________________________ devel-announce mailing list -- devel-annou...@lists.fedoraproject.org To unsubscribe send an email to devel-announce-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel-annou...@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue -- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
