On Tuesday, January 26, 2021 9:32 AM, Johnny Hughes <johnny at centos.org> wrote: > On 1/25/21 10:21 AM, Gena Makhomed wrote: > > > On 14.01.2021 23:06, Brian Stinson wrote: > > > > > On Thu, Jan 14, 2021, at 14:50, Gena Makhomed wrote: > > > > > > I found regression in the CentOS Stream 8 kernels, and create > > > > bug report https://bugzilla.redhat.com/show_bug.cgi?id=1913806 > > > > After reading https://wiki.centos.org/Contribute/CentOSStream > > > > I try to download kernel sources and look for root cause of bug, > > > > but dist-git and source-git repos does not contain kernel sources, > > > > even if I try to execute %prep phase. For this reason I can't > > > > even try to find root cause of this bug by myself. > > > > This bug is critical for me, because currently I try to use > > > > systemd-nspawn for containers virtualization in production. > > > > If future CentOS 8.4 will contain this bug - it will be disaster. > > > > What else I can do (as user or developer) for fixing this bug? > > > > > We should have a new kernel out in the next couple of days, next step > > > would be to see if that bug is fixed in this build: > > > https://koji.mbox.centos.org/koji/buildinfo?buildID=15499 > > > > I check latest CentOS Stream 8 kernel version 4.18.0-269.el8.x86_64 > > > > - this regression with broken nspawn containers still present here. > > > > Does any chance exist that this kernel regression will be fixed > > in CentOS Stream 8 before CentOS 8.4 / RHEL 8.4 will be released? > > > > > One thing we might be able to update about the documentation if you're > > > looking into the sources is to add a link here: > > > https://wiki.centos.org/Sources (see the section there describing the > > > get_sources.sh script). We store large sources in a separate > > > lookaside, not directly in the dist-git trees. > > > > Thank you! Now I can download source - but all patches are merged > > into one file linux-4.18.0-269.el8.tar.xz - I have no chances > > to find which patch to the kernel source tree causes this regression > > with broken systemd-nspawn containers. As I understand, only Red Hat > > kernel developers now can find the root cause of regression and fix it. > > This is how Red Hat releases all kernel sources. > > > And all I can do is file a bug > > https://bugzilla.redhat.com/show_bug.cgi?id=1913806 > > and hope for the best? It is true that this has been how Red Hat releases all it's kernel sources and the issue has been outside of CentOS' control. The immediate answer would be to file a bug and hope for the best (I share your frustration). However, this issue has been heard and is being addressed already. Check the thread "RFC: Stream Kernel SIG Proposal." On Jan 20th, Mike McGrath indicates part of the move of Stream into gitlab in a few weeks includes being able to see individual git commits. At this point I think he would have committed to a more solid date if he could have. My best guess is there are issues going with with the gitlab transition outside of his control. Maybe we will learn more from the CentOS Dojo @FOSDEM.