https://pad.riseup.net/p/ReleaseEngineering-20210331 is a backup etherpad as we can't really use etherpad.opensuse.org because of https://progress.opensuse.org/issues/90230 All meeting minutes can be found here: https://etherpad.opensuse.org/p/ReleaseEngineering-meeting ## Participants lkocman, rbrown,guillaumeg, adrian, nathankeith, ddemaio, jackson, sbahling, maxlin, gerald, adrians, sarah, wolfgang !!! Please be aware that this meeting is hosted on https://meet.opensuse.org/ReleaseEngineeringMeeting ## Leap Build failures x86_64 20/7 (today 11/1) i586 0/0 (0/0) aarch64 74/10 (74/9) armv7 117/99 113/139 ppc64le 119/115 (112/113) s390x 163/625 (160/624) Agreement is to have 0 build failures on intel, and hopefully aarch64. Everything else should be as good as it can get. https://progress.opensuse.org/issues/76348 Discussion around noarch source https://bugzilla.opensuse.org/show_bug.cgi?id=1183097 We'd like to discuss proposals on the bottom of the. Looking with Jiri Srain into free variant of SLE Micro, something that could be migratable to SLE, otherwise it would be Kubic. April 21st is RC Code submissiond deadline. I'll work on an assignment and deadlines for Leap 15.3 RC tasks. Still have to look into what pieces of translations need to be handled after CtLG and what not. I did briefly talk to sbrabec on this topic. A new Leap 15.3 build today Build117.13, hopefully we can get it published soon. ## openSUSE Tumbleweed * SELinux 3.4 update caused a larger rebuild than expected, issues for users were saved thanks to Bernhard's excellent work delaying publishing of large snapshots (Thanks again!) * SELinux enabling during the installation phase was improved, cutting down the reboots required * GNOME 40 is submitted, waiting for some fixes in other packages impacted by the glib2 changes * Python 3.9 is in staging, requiring some modules to be improved * GCC 11 as the default compiler is being tested in a Staging ## Richard (Kubic/TW MicroOS) MicroOS Desktop (GNOME) Beta has been released, some pattern tuning identified before it can be considered production ready transactional-updates 3.3.0 was released SELinux is now the default on MicroOS ## Max * Hackweek, Leap has not making much progress * patterns-cinnamon is now exist in Leap 15.3(Note: not Backports) * arm and powerpc have got problem with plymouth, we've a reverted plymouth in Leap for the time being bsc#1184087 ## Guillaume - Arm Tumbleweed: * Rolling, no blocker * In-kernel Memory Tagging Extension (MTE) support via KASAN for aarch64 will go to kernel-debug. Leap 15.3 ARM: * aarch64: * Plymouth bug: https://bugzilla.suse.com/show_bug.cgi?id=1184087 (update reverted by Max) * ghc: unresolvable: https://bugzilla.opensuse.org/show_bug.cgi?id=1184126 * Few updates submitted to backports to get rid of unresolvables * Armv7: * initial openQA setup done : https://openqa.opensuse.org/tests/overview?distri=opensuse&version=15.3&build=2.15&groupid=79 * armv7 NET ISO is now bootable but grub has no theme and it hangs at 'Hardware detection'. Let's see if next snapshot is better. WSL: Stalled: WIP to have WSL on aarch64 on https://build.opensuse.org/project/show/home:Guillaume_G:WSL The goal is to have WSL-DistroLauncher package built for aarch64. Help welcomed. SUSE is currently interviewing candidates for the WSL Engineering post. So far Jeff is still the only person working on the Engineering efforts (native kiwi support, and some current WSL build blockers on SLE). Lubos is currently considering a hackweek project for WSL / aarch64. This is currently the plan. ## Michel ppc64le Not available ## Sarah - s390x * openSUSE for s390x is maintained by the community in the future. There is a contract rule, that no IBMer is allowed to contibute for openSUSE on IBM Z without having anybody for Fedora in the same role. * Tumbleweed: no blockers * Leap 15.3 test issue because of SLE test dependency: https://progress.opensuse.org/issues/81685 (will be ignored) Lubos: Contact Ihno Kumreich for any questions for s390x/SLE Engineering. I can try to help as well. Will not be available next meeting. ## Doug * Leap 15.3 * Almost finished with https://en.opensuse.org/Features_15.3 * Beta t-shirts still coming in (shipped a few in NA) * Marketing meeting and sprints today and tomorrow - https://bit.ly/2OeuD8c * openSUSE Conference (June 18 - 20) * GSoC Mentor Org (student applications until April 13) * Leap 15.3 Box - discussion in progress (currently just missing up2date German install manual) ## Dirk Not much progress last week due to Hackweek. Resumed looking at Leap 15.3/Step 15 work. A lot of issues with backports and missing updates in SLE side. Step: Still a lot of unsolved issues (e.g. requirements on sles- release, but sles-release disabled for non-SLE architectures) ## Gerald Based on report on the project@ list: can we see to spread out Beta/RC releases of Leap and larger Tumbleweed snapshots as not to overload mirrors? Lubos indicated he'd look into that with Dominique. ## Wolfgang (Package Hub), Scott Bahling Issue with noarch packages, mentioned at the end of meeting notes. (Possible reworking of PH in 15.4) ## Adrian - CtLG or openSUSE Step RISC-V/TW is not represented on this call. We still don't know how to release ftp-tree for armv7. Lubos: https://github.com/openSUSE/get-o-o/issues/34 (Inputs from Max, Wolfgang, Scott, Lubos). A discussion regarding https://bugzilla.opensuse.org/show_bug.cgi?id=1183097 We'd like to have an approval from Adrian on which solution to go for. Currently Preferred solution is (4). Options: 1) we delete(or disable flavor) every package if it builds noarch(either it has noarch binary only or containing arch-dependent + noarch[1]) binary RPM and that package has been excluding to build on any arch. Take mumps as the example, we should disable gnu-mpich- hpc/gnu-mvapich2-hpc/gnu-mvapich3-hpc/ gnu-mvapich4-hpc flavor completely, leave other flavor as it is. Another example librepository[2], just delete it since it’s noarch only(BuildArch: noarch) package. Pro: solve noarch binary issue Con: PackageHub would not providing some packages for all archs. *noarch* package is fine, but *arch-dependent + noarch* will losing since the package removal, for example libmumps-gnu-openmpi4-hpc-5.3.5- bp153.1.1.ppc64le.rpm in [1] This may increase amount of open bugs from customer side such as “Package is no longer available in Backports”. [1] https://build.opensuse.org/package/binaries/openSUSE:Backports:SLE-15-SP3/mu... see ppc64le result [2] https://build.opensuse.org/package/show/openSUSE:Backports:SLE-15-SP3/librep... 2) somehow filter out these duplicite noarch rpms on publishing/releasing process Con: this is would be a manually maintained list, which we’d like to see disappear in 15.4. But will it? Pro: PackageHub consumers will still get all rpms unlike in (1) 3) Long term, handle building exclusively in Leap / SLE projects, backports would just then handle publishing of “diff”. Right now we have to build same srpms twice and figure out the magic. This should be something to look into for 15.4 Con: Quite a lot of work in the beginning Pro: cleaner build and release pipeline (this issue will simply go away) 4) like 1) we delete(or disable flavor) every package if it builds noarch(either it has noarch binary only or containing arch-dependent + noarch[1]) binary RPM and that package has been excluding to build on any arch. Proving them from subpackage module (repository for unrelease packages from SLE) instead(SLE built). Pro: solve noarch binary issue. And those binaries in subpackage module were SLE built, not the Backports rebuild binaries. Con: find a list of packages adding to subpackage module Adrian: 4th seems to be the option if it's also Runtime issues. Adrian: we have to create an additional channel (subpackages already exist, we just need to modify it). Max: will work on the list of packages that will have to be added to the subpackage module. (5)Adrian: we could add a side-layer on top of current backpots to incorporate SLE packages and opt out Leap.