During installation of the latest kernel (5.6.0-0.rc3.git2.2.fc33.x86_64) there is a post-install step that takes at least 25 minutes to run (actually I'm still waiting for it to finish). 2145 pts/0 S+ 0:00 | \_ sudo dnf update --best glibc kernel 2147 pts/0 S+ 0:14 | \_ /usr/bin/python3 /usr/bin/dnf update --best glibc kernel 2211 pts/0 S+ 0:00 | \_ /bin/sh /var/tmp/rpm-tmp.zOSxPy 3 2212 pts/0 S+ 0:28 | \_ /usr/bin/bash /usr/sbin/weak-modules --add-kernel 5.6.0-0.rc3.git2.2.fc33.x86_64 63587 pts/0 R+ 0:02 | \_ /sbin/depmod -C /tmp/weak-modules.0Ig9Wa/depmod.conf -naeE /tmp/weak-modules.0Ig9Wa/symvers-5.6.0-0.rc3.git2.2.fc33.x86_64 5.6.0-0.rc3.git2.2.fc33.x86_64 63588 pts/0 S+ 0:00 | \_ grep /tmp/weak-modules.0Ig9Wa/5.6.0-0.rc3.git2.2.fc33.x86_64/weak-updates This is on a very fast AMD machine.
FWIW, here(Ryzen 7 2700X) it finished after 11 minutes, but this is with a working depmod (kmod-26) instead of the slightly broken one (kmod-27), bug 1808430)
I was using kmod-26-4.fc31.x86_64. IMHO even 11 minutes is much too long.
Yeah, I noticed this here as well... I'm not sure how long it took, but many minutes...
Gah. I've backed the weak-modules change out again. 5.6.0-0.rc4.git0 should behave properly. Sorry.
(In reply to Richard W.M. Jones from comment #2) > I was using kmod-26-4.fc31.x86_64. IMHO even 11 minutes is much too long. Richard, could you help me to find the root cause, if it's not https://bugzilla.redhat.com/show_bug.cgi?id=1817581 ? I need more detailed information about the setup -- kernels and modules, especially in extra/ dirs.
Seems very different to me. In any case you will need to ask someone on the kernel team in order to get a definitive answer.
This bug seems to be back with kernel-core-5.7.0-0.rc5.1.fc33.x86_64 It's been running the weak modules thing for 20+ minutes here.
74058 pts/3 S+ 0:00 \_ sudo dnf reinstall kernel-core 74068 pts/3 S+ 0:16 \_ /usr/bin/python3 /bin/dnf reinstall kernel-core 83518 pts/3 S+ 0:00 \_ /bin/sh /var/tmp/rpm-tmp.y0PrWR 3 83519 pts/3 S+ 0:01 \_ /usr/bin/bash /usr/sbin/weak-modules --add-kernel 5.7.0-0.rc5.1.fc33.x86_64 198035 pts/3 R+ 0:01 \_ /sbin/depmod -C /tmp/weak-modules.fRWcad/depmod.conf -naeE /tmp/weak-mo 198036 pts/3 S+ 0:00 \_ grep /tmp/weak-modules.fRWcad/5.7.0-0.rc5.1.fc33.x86_64/weak-updates
So the root cause was the kernel-modules-extra package was installing modules to the wrong location. This has since been fixed, but if you've got old kernel-modules-extra packages sitting around the weak-modules script continues to pick them up. A workaround should be to remove all the old kernel-modules-extra packages you have installed.
I don't believe I had it installed. It's not installed now (after the upgrade finally finished): $ rpm -q kernel-modules-extra package kernel-modules-extra is not installed and it's not mentioned in /var/log/dnf.log*: $ sudo grep kernel-modules-extra /var/log/dnf.log* [no output] So I guess it hasn't been installed for a while (these logs go back to 2019-07). Could there be any other package which would cause this?
Check contents of /lib/modules/*/extra dirs. Time is basically O(m * k) where m is amount of modules in extra/'s and k is number of installed kernels.
$ ls /lib/modules/*/extra /lib/modules/5.2.0-0.rc3.git3.2.fc31.x86_64+debug/extra: drivers fs net /lib/modules/5.6.0-0.rc3.git0.1.fc32.x86_64+debug/extra: drivers fs net /lib/modules/5.7.0-0.rc5.1.fc33.x86_64+debug/extra: (last one seems to be empty) $ find /lib/modules/*/extra -type f | wc -l 271 Anyway I'll close this one again, thanks for the explanation.
*** Bug 1851216 has been marked as a duplicate of this bug. ***