Description of problem: AArch64 disk images include '/etc/sysconfig/network-scripts/ifcfg-enp1s0'. This configuration file should be removed from the image after creation to avoid conflicts. cat ./etc/sysconfig/network-scripts/ifcfg-enp1s0 # Generated by dracut initrd NAME="enp1s0" DEVICE="enp1s0" ONBOOT="yes" NETBOOT="yes" UUID="1f4fa52a-765b-467d-be98-2ac3fdb49490" IPV6INIT="yes" BOOTPROTO="dhcp" TYPE="Ethernet" PROXY_METHOD="none" BROWSER_ONLY="no" DEFROUTE="yes" IPV4_FAILURE_FATAL="no" IPV6_AUTOCONF="yes" IPV6_DEFROUTE="yes" IPV6_FAILURE_FATAL="no" Version-Release number of selected component (if applicable): Latest disk images in F27.
https://pagure.io/fedora-kickstarts/pull-request/313
Proposed FE for final.
This sounds worth an FE, though I'm *slightly* worried that the change will apply to all disk images (some of which are release-blocking). It looks both safe and correct (until the name of the interface changes due to some change in ImageFactory or systemd or kernel, of course...), but there's always the chance...
(In reply to Adam Williamson from comment #3) > This sounds worth an FE, though I'm *slightly* worried that the change will > apply to all disk images (some of which are release-blocking). It looks both It doesn't affect anything except aarch64 images, if you grep for the includes you'll see the only included ones are used currently for aarch64: fedora-disk-base.ks:# fedora-disk-base.ks fedora-disk-minimal.ks:%include fedora-disk-base.ks fedora-disk-server.ks:%include fedora-disk-base.ks fedora-disk-workstation.ks:%include fedora-disk-base.ks > safe and correct (until the name of the interface changes due to some change > in ImageFactory or systemd or kernel, of course...), but there's always the > chance... Sure, but it actually doesn't just come down to interface name (which is static due to PCI buses/slots) but also the HWADDR generated by libvirt/qemu for the virtual NIC, and yes these happen in a range allowed for allocation but it's extremely unlikely, even if a device (and there are somee) has a pci attached NIC on BUS 1 slot 0, that it'll have the same MAC address of that which was automatically generated and network manager/initial-setup can recreate the file. All those words to justify I feel this is a safe operation :)
"if you grep for the includes you'll see the only included ones are used currently for aarch64:" but they're not *only* used for aarch64, are they? they're also used for generating the cloud and 32-bit ARM disk images, aren't they?
(In reply to Adam Williamson from comment #5) > "if you grep for the includes you'll see the only included ones are used > currently for aarch64:" > > but they're not *only* used for aarch64, are they? they're also used for > generating the cloud and 32-bit ARM disk images, aren't they? Yes they *only* used in <= F27 (and currently rawhide but this is out of scope for this bug): [perobins@neo fedora-kickstarts (f27 %)]$ grep fedora-disk-base.ks * grep: custom: Is a directory fedora-disk-base.ks:# fedora-disk-base.ks fedora-disk-minimal.ks:%include fedora-disk-base.ks fedora-disk-server.ks:%include fedora-disk-base.ks fedora-disk-workstation.ks:%include fedora-disk-base.ks grep: l10n: Is a directory grep: snippets: Is a directory grep: templates: Is a directory grep: tools: Is a directory [perobins@neo fedora-kickstarts (f27 %)]$ Then if you reference the fedora pungi config: https://pagure.io/pungi-fedora/blob/f27/f/fedora.conf#_405 You'll see that fedora-disk* is only referenced with aarch64 users
Discussed during the 2017-11-06 blocker review meeting: [1] The decision to classify this bug as an AcceptedFreezeException was made as the bug cannot be fixed with an update and only affects aarch64 images. [1] https://meetbot.fedoraproject.org/fedora-blocker-review/2017-11-06/f27-blocker-review.2017-11-06-17.00.txt
Well, this was granted an FE but the PR was never merged, apparently...what's the status?
(In reply to Adam Williamson from comment #8) > Well, this was granted an FE but the PR was never merged, > apparently...what's the status? Not sure who was meant to merge it. I'll just add it now as a common bugs entry so it's documented.