Hey folks! Just to keep everyone in the loop regarding F39 plans.
As you may have noticed, we've slipped once or twice already (depending on whether you count the "early target date") and are in danger of slipping again. The go/no-go meeting is scheduled for tomorrow (2023- 10-26). The outstanding blockers are all Raspberry Pi-related, aside from the shim one we've been waiving for several releases and intend to waive again.
Matthew Miller, Kevin Fenzi and I came up with this plan: we're going to run a compose right now without fixes for the two outstanding Pi blockers (2241252 and 2244305). If QA can get sufficient testing on this done by the go/no-go meeting, we can discuss the possibility of shipping it and noting that there are known issues with Raspberry Pi that are taking time to resolve, and Pi users should not install or upgrade to F39 until they're resolved (or something like that).
If at any point a fix for 2241252 shows up, we'll run another compose with the fix included. If that gets sufficient testing by the time of the meeting, we can also consider that as a candidate to ship. If ARM team decide to attempt a fix for 2244305 we'd also pull that in, but if not, we think it's reasonable to consider revoting or waiving that bug, as it seems not to happen very commonly or consistently and the proposed "fix" apparently comes with tradeoffs of some kind.
So, QA folks, please stand ready to test one or two candidate composes soon. If we wind up with two, we will consider most test results to apply to both, as the only difference should be uboot-tools; we would want to run ARM hardware tests, at least, on both composes if possible. The usual announcement mails will be sent for the completed composes.
Thanks folks!
On Wed, 2023-10-25 at 11:23 -0700, Adam Williamson wrote:
Hey folks! Just to keep everyone in the loop regarding F39 plans.
As you may have noticed, we've slipped once or twice already (depending on whether you count the "early target date") and are in danger of slipping again. The go/no-go meeting is scheduled for tomorrow (2023- 10-26). The outstanding blockers are all Raspberry Pi-related, aside from the shim one we've been waiving for several releases and intend to waive again.
Matthew Miller, Kevin Fenzi and I came up with this plan: we're going to run a compose right now without fixes for the two outstanding Pi blockers (2241252 and 2244305). If QA can get sufficient testing on this done by the go/no-go meeting, we can discuss the possibility of shipping it and noting that there are known issues with Raspberry Pi that are taking time to resolve, and Pi users should not install or upgrade to F39 until they're resolved (or something like that).
If at any point a fix for 2241252 shows up, we'll run another compose with the fix included. If that gets sufficient testing by the time of the meeting, we can also consider that as a candidate to ship. If ARM team decide to attempt a fix for 2244305 we'd also pull that in, but if not, we think it's reasonable to consider revoting or waiving that bug, as it seems not to happen very commonly or consistently and the proposed "fix" apparently comes with tradeoffs of some kind.
So, QA folks, please stand ready to test one or two candidate composes soon. If we wind up with two, we will consider most test results to apply to both, as the only difference should be uboot-tools; we would want to run ARM hardware tests, at least, on both composes if possible. The usual announcement mails will be sent for the completed composes.
Thanks folks!
Update on this: the first candidate without Pi fixes is done and currently available for testing - see https://fedoraproject.org/wiki/Test_Results:Fedora_39_RC_1.1_Summary . The second candidate is running, when it is done, https://fedoraproject.org/wiki/Test_Results:Fedora_39_RC_1.2_Summary will be live.
Most tests of either candidate will be valid for both, but we especially would like testing of the second candidate (when it's done) on ARM hardware - obviously on Raspberry Pi, but also on any other ARM hardware folks have lying around. We ended up having to revert uboot- tools to an older version to try and address the Pi issues, so we need to check that hasn't broken anything else important. If you run into problems, please file a bug, propose it as a release blocker, and maybe reply here just to be sure :)
Thanks!
-
Adam Williamson