Just to provide a heads-up: the corresponding changes are now live and patch submitters should receive email notifications moving forward. Regards, Daniel (on behalf of the Kernel CI team at Meta) On Wed, Nov 15, 2023 at 12:45:46AM +0000, Daniel Müller wrote: > Hi everyone, > > As I am sure most of you are aware, BPF has a CI system that runs on all patch > submissions. Details have been presented at various conferences, with several > recordings and slide decks detailing workings available (e.g., [0]). In the past > this system has, in large parts, been a tool for maintainers to quickly spot > problems, as results get bubbled up to Patchwork [1]. > > We believe that it is equally important for patch submitters to be made aware of > CI results. To that end, we added support for email notifications to the CI > system a while back. For some time now, these notifications had been enabled for > chosen Meta BPF developers as well as BPF maintainers to gather and address > feedback. > > At this point, we would like to take the next step and have notification emails > go out to *all* patch submitters. > > # What to expect > > Everybody submitting a patch will be informed about success or failure of said > submission once CI concluded or a failure has been detected via email. A link to > the GitHub Actions run will be included (e.g., [2]), which allows for quick > navigation to the individual failed test runs in case of failure. > > All emails originate from bot+bpf-ci@xxxxxxxxxx, allowing for easy filtering. > > Note that as per our current CI design, patches are effectively tested > continuously. That is, if an update is pushed to the upstream branch (bpf or > bpf-next), the patch will be rebased and retested. An email notification will > only be sent if the current result is different from what it was before (i.e., > success -> failure or failure -> success). New patch versions (e.g., update from > v1 -> v2) will always trigger a new email initially. > > Please also note that while the Kernel CI team at Meta tries to address > infrastructure issues affecting the CI system in a timely manner, we do not have > the resources to fix all sources of selftest flakiness. We encourage everyone to > send out fixes addressing problems, including when not directly caused by their > submission. It is in this context that submissions such as Tao's recent > test_maps fix [3] are very valuable and will only become more so as test > flakiness has the potential to affect more people (thanks!). > > # Moving forward > > If you have any concerns or want to be excluded from CI emails for your > submissions, please reach out to kernel-ci@xxxxxxxx. If not, expect to receive > email notifications for your own patch submissions starting some time next week. > Once enabled, please feel free to share any feedback you have with > kernel-ci@xxxxxxxx. Eventually, our intent is to CC the bpf mailing list on > all such CI emails. > > Regards, > Daniel (on behalf of the Kernel CI team at Meta) > > [0] http://vger.kernel.org/bpfconf2022_material/lsfmmbpf2022-bpf-ci.pdf > [1] https://patchwork.kernel.org/project/netdevbpf/list/ > [2] https://github.com/kernel-patches/bpf/actions/runs/6867078893 > [3] https://lore.kernel.org/bpf/20231101032455.3808547-1-houtao@xxxxxxxxxxxxxxx/ >
