Description of problem: $ dnf list openshot --enablerepo=*testing RPM Fusion for Fedora 25 - Nonfree - Test Updates 26 kB/s | 1.2 kB 00:00 Fedora 25 - x86_64 - Test Updates 4.2 MB/s | 5.1 MB 00:01 system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. Aborted (core dumped) [daddy@arcturus VOICE]$ Version-Release number of selected component: system-python-3.5.3-6.fc25 Additional info: reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /usr/bin/dnf list openshot --enablerepo=*testing crash_function: load_ext executable: /usr/libexec/system-python global_pid: 18150 kernel: 4.11.5-200.fc25.x86_64 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project runlevel: N 5 type: CCpp uid: 1001 Truncated backtrace: Thread no. 1 (10 frames) #4 load_ext at /usr/src/debug/hawkey-0.6.4/src/sack.c:354 #5 hy_sack_load_repo at /usr/src/debug/hawkey-0.6.4/src/sack.c:1049 #6 load_repo at /usr/src/debug/hawkey-0.6.4/src/python/sack-py.c:455 #7 PyCFunction_Call at /usr/src/debug/Python-3.5.3/Objects/methodobject.c:98 #8 do_call at /usr/src/debug/Python-3.5.3/Python/ceval.c:5006 #9 call_function at /usr/src/debug/Python-3.5.3/Python/ceval.c:4805 #10 PyEval_EvalFrameEx at /usr/src/debug/Python-3.5.3/Python/ceval.c:3304 #11 fast_function at /usr/src/debug/Python-3.5.3/Python/ceval.c:4876 #12 call_function at /usr/src/debug/Python-3.5.3/Python/ceval.c:4803 #13 PyEval_EvalFrameEx at /usr/src/debug/Python-3.5.3/Python/ceval.c:3304
Created attachment 1289972 [details] File: backtrace
Created attachment 1289973 [details] File: cgroup
Created attachment 1289974 [details] File: core_backtrace
Created attachment 1289975 [details] File: dnf.txt
Created attachment 1289977 [details] File: dso_list
Created attachment 1289978 [details] File: environ
Created attachment 1289979 [details] File: limits
Created attachment 1289980 [details] File: maps
Created attachment 1289981 [details] File: mountinfo
Created attachment 1289982 [details] File: namespaces
Created attachment 1289983 [details] File: open_fds
Created attachment 1289984 [details] File: proc_pid_status
Created attachment 1289985 [details] File: var_log_messages
The dnf.txt file is an strace -o output file.
Similar problem has been detected: [vsapronenko@localhost ~]$ dnf info flash-player RPM Fusion for Fedora 25 - Nonfree - Updates 303 B/s | 37 kB 02:05 Fedora 25 - x86_64 - Updates 867 kB/s | 23 MB 00:27 RPM Fusion for Fedora 25 - Nonfree 1.1 kB/s | 144 kB 02:05 Fedora 25 - x86_64 2.1 MB/s | 50 MB 00:23 RPM Fusion for Fedora 25 - Free 96 kB/s | 515 kB 00:05 Fedora 25 - x86_64 - Test Updates 1.4 MB/s | 5.1 MB 00:03 system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. Aborted (збережено знімок оперативної пам’яті) [vsapronenko@localhost ~]$ dnf info flash-player Adobe Systems Incorporated 385 B/s | 1.9 kB 00:05 RPM Fusion for Fedora 25 - Free - Updates 916 kB/s | 275 kB 00:00 Copr repo for haguichi owned by ztefn 4.1 kB/s | 2.5 kB 00:00 google-chrome 761 B/s | 3.8 kB 00:05 PostgreSQL 9.5 25 - x86_64 31 kB/s | 313 kB 00:09 Repository for Fedy 4.8 kB/s | 8.1 kB 00:01 system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. Aborted (збережено знімок оперативної пам’яті) reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /usr/bin/dnf info flash-player crash_function: load_ext executable: /usr/libexec/system-python global_pid: 11067 kernel: 4.11.5-200.fc25.x86_64 package: system-python-3.5.3-6.fc25 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project reason: system-python killed by SIGABRT runlevel: N 5 type: CCpp uid: 1000
Similar problem has been detected: [vsapronenko@localhost ~]$ sudo dnf upgrade [sudo] пароль до vsapronenko: Copr repo for PyCharm owned by phracek 62 kB/s | 60 kB 00:00 PostgreSQL 9.5 25 - x86_64 49 kB/s | 313 kB 00:06 system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. Aborted [vsapronenko@localhost ~]$ sudo dnf upgrade Fedora 25 - x86_64 - VirtualBox 6.3 kB/s | 33 kB 00:05 system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. Aborted reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /bin/dnf upgrade crash_function: load_ext executable: /usr/libexec/system-python global_pid: 12145 kernel: 4.11.5-200.fc25.x86_64 package: system-python-3.5.3-6.fc25 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project reason: system-python killed by SIGABRT runlevel: N 5 type: CCpp uid: 0
Similar problem has been detected: This happened for the first time when I ran "sudo dnf distro-sync", but it seems to happen for every dnf command ("dnf list installed", "dnf repolist", etc). Esentially I can't do anything with dnf. reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /usr/bin/dnf distro-sync crash_function: load_ext executable: /usr/libexec/system-python global_pid: 32366 kernel: 4.11.5-200.fc25.x86_64 package: system-python-3.5.3-6.fc25 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project reason: system-python killed by SIGABRT runlevel: N 5 type: CCpp uid: 0
This seems to be the same as bug 1358885. This is likely not a core python crash but dnf or hawkey crash, reassigning.
There's some problematic package in updates-testing. Dnf crashes after updates-testing metadata is loaded. Look at this: # rm /var/cache/dnf/* -rf # dnf repolist --disablerepo='*' --enablerepo=fedora Fedora 25 - x86_64 8.1 MB/s | 50 MB 00:06 Last metadata expiration check: 0:00:17 ago on Wed Jun 21 13:04:48 2017. repo id repo name status *fedora Fedora 25 - x86_64 51,669 # dnf repolist --disablerepo='*' --enablerepo=fedora --enablerepo=updates Fedora 25 - x86_64 - Updates 10 MB/s | 23 MB 00:02 Last metadata expiration check: 0:00:06 ago on Wed Jun 21 13:05:14 2017. repo id repo name status *fedora Fedora 25 - x86_64 51,669 *updates Fedora 25 - x86_64 - Updates 20,129 # dnf repolist --disablerepo='*' --enablerepo=fedora --enablerepo=updates --enablerepo=updates-testing Fedora 25 - x86_64 - Test Updates 5.6 MB/s | 5.1 MB 00:00 system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. Aborted (core dumped)
We need to identify which package from updates-testing causes this, so that we can unpush it (also make sure it doesn't get pushed stable!). And then we need the bug fixed.
*** Bug 1463505 has been marked as a duplicate of this bug. ***
*** Bug 1463582 has been marked as a duplicate of this bug. ***
*** Bug 1463580 has been marked as a duplicate of this bug. ***
Kamil, For me the bug occured today about 9 AM GMT, yesterday aboyt 9 PM GMT everything was OK. Check what was pushed to updates-testing in this particular period of time
This is also happening on Fedora 24 (with updates-testing) with: dnf-1.1.10-4.fc24.noarch hawkey-0.6.4-3.fc24.x86_64 python2-hawkey-0.6.4-3.fc24.x86_64 python3-dnf-1.1.10-4.fc24.noarch python3-hawkey-0.6.4-3.fc24.x86_64 system-python-3.5.3-6.fc24.x86_64
Doesn't seem to happen with F26 with: dnf-2.5.1-1.fc26.noarch libdnf-0.9.1-1.fc26.x86_64 python3-dnf-2.5.1-1.fc26.noarch python3-hawkey-0.9.1-1.fc26.x86_64 system-python-3.6.1-6.fc26.x86_64 Probably either because dnf 2 fixes this, or because the offending package is not in F26 updates-testing repo.
As far as I know dnf 2 does not include hawkey and rather uses libdnf
It is irrelevant here..
https://dl.fedoraproject.org/pub/fedora/linux/updates/testing/25/x86_64/repodata/ Contains 2 *-updateinfo.xml.xz... and one is referenced from repomd.xml is actually sqlite DB... So something fucked up mashing or something like that.. Not a bug in DNF.
*** Bug 1463645 has been marked as a duplicate of this bug. ***
Has anyone notified infrastructure?
I have spotted this issue today for the first time when running 'dnf install ..' after having installed the latest packages from "updates-testing" via gnome-software. Yesterday, 'dnf' was still working correctly.
We have identified the issue and Infra is working on fixing the repos. Everybody please just wait, it should be fixed in a day or less.
(In reply to Igor Gnatenko from comment #29) > Contains 2 *-updateinfo.xml.xz... and one is referenced from repomd.xml is > actually sqlite DB... <snip> > Not a bug in DNF. It would still be very nice if whatever is crashing instead handled the error and reported something actionable, like “parsing file downloaded from $URL failed”.
@mtrmac: Newer libdnf already does set an error properly in this case, it's only f25's dnf that has this problem.
*** Bug 1463689 has been marked as a duplicate of this bug. ***
Hitting this too, thanks for looking into this quickly. Are there any suggested workarounds at the moment?
Encountering this on Fedora 25, updated yesterday, and note that I *DO NOT* have updates-testing enabled.
Yes, it seems to affect all repos, not just updates-testing. It depends how up-to-date your mirror is. Please wait. You might be able to use packagekit/gnome-software to install packages as a workaround (not tested).
(In reply to Kamil Páral from comment #20) > We need to identify which package from updates-testing causes this, so that > we can unpush it (also make sure it doesn't get pushed stable!). And then we > need the bug fixed. Too late: $ sudo dnf repolist --disablerepo='*' --enablerepo=fedora [sudo] password for garry: Fedora 25 - x86_64 1.2 MB/s | 50 MB 00:43 A Last metadata expiration check: 0:00:41 ago on Wed Jun 21 11:25:23 2017. repo id repo name status *fedora Fedora 25 - x86_64 51,669 $ sudo dnf repolist --disablerepo='*' --enablerepo=fedora --enablerepo=updates system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. zsh: abort sudo dnf repolist --disablerepo='*' --enablerepo=fedora --enablerepo=updates $
(In reply to Kamil Páral from comment #33) > We have identified the issue and Infra is working on fixing the repos. > Everybody please just wait, it should be fixed in a day or less. Thanks for your support!
since today got that too: system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: Assertion `ret == 0' failed. I did: rm -f /var/lib/rpm/__db* rpm --rebuilddb dnf clean al and everything works fine again.
(In reply to Donald O. from comment #42) > since today got that too: > system-python: /builddir/build/BUILD/hawkey-0.6.4/src/sack.c:354: load_ext: > Assertion `ret == 0' failed. > > I did: > rm -f /var/lib/rpm/__db* > rpm --rebuilddb > dnf clean al > > and everything works fine again. It *looks like* it works, because it starts downloading repos, but once it actually goes into the phase where it computes the list of updates, it cores again for me.
Just to let you know: In my F25 there are 2 repos causing the problem: updates-testing and updates. If I disable both, dnf works...
In one case, dnf clean all worked For another, From https://bugzilla.redhat.com/show_bug.cgi?id=1463561#c44, I removed contents of /var/cache/yum/x86_64/25/updates and it started working for me.
I also did rm -f /var/lib/rpm/__db* rpm --rebuilddb dnf clean all works ok except when reaching repo Fedora 25 - x86_64 - Updates Disabling Fedora 25 - x86_64 - Updates dnf works..
I am on Fedora 24 still. Have the same issue. Updates-Testing is not enabled. Disabling Fedora-Updates allows dnf to complete.
Found my solution at: http://forums.fedoraforum.org/showthread.php?p=1789198 ------- This worked for me ... (although on a second machine I had to do it twice ,,,) # dnf clean all # dnf --disablerepo=updates update # dnf update ----- After above my dnf no longer gives the described error message. If this is any help: $ uname -a Linux pilot.local 4.11.5-200.fc25.x86_64 #1 SMP Wed Jun 14 17:17:29 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
The master mirrors have been fixed, and mirrormanager has been told to not hand out the previous (bad) checksum. So, dnf --refresh or dnf clean all should get everyone working now provided you are using the (default) metalink. Other mirrors should sync up over today.
Confirmed that fixed it for me.
Thank you for the quick response on this! It seems to be fixed now
Thanks a lot to Patrick Uiterwijk for quickfixing it and everyone who was involved ;)
*** Bug 1463751 has been marked as a duplicate of this bug. ***
This Bodhi issue has been filed about removing the code that caused this problem, and it references a PR that will fix it: https://github.com/fedora-infra/bodhi/issues/1634
got resonse from other users who also can fix the problem by my recipe. Therefore I conclude the problem is related to the well-known https://fedoramagazine.org/psa-errors-updating-libdb/
(In reply to Donald O. from comment #56) > got resonse from other users who also can fix the problem by my recipe. A success of your recipe depends on dnf getting either "sane" data for updates repo or none at all. As fixed repos are propagating through mirrors chances are that you will end up OK. OTOH if you had dnf-makecache.timer active then every time it fired with "bad" repos dnf was dumping a core and this was quite quickly accumulating in /var/lib/systemd/coredump/. It is a good idea to check what it sitting there.
Similar problem has been detected: reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /usr/bin/dnf info flash-player crash_function: load_ext executable: /usr/libexec/system-python global_pid: 4107 kernel: 4.11.5-200.fc25.x86_64 package: system-python-3.5.3-6.fc25 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project reason: system-python killed by SIGABRT runlevel: N 5 type: CCpp uid: 1000
*** Bug 1464344 has been marked as a duplicate of this bug. ***
Similar problem has been detected: reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /usr/bin/dnf info flash-plugin crash_function: load_ext executable: /usr/libexec/system-python global_pid: 25016 kernel: 4.11.6-200.fc25.x86_64 package: system-python-3.5.3-6.fc25 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project reason: system-python killed by SIGABRT runlevel: N 5 type: CCpp uid: 1000
Similar problem has been detected: Didn't know the problem had occurred. Found it in the problem reporting tool reporter: libreport-2.8.0 backtrace_rating: 4 cmdline: /usr/libexec/system-python /usr/bin/dnf makecache timer crash_function: load_ext executable: /usr/libexec/system-python global_pid: 8077 kernel: 4.11.5-200.fc25.x86_64 package: system-python-3.5.3-6.fc25 pkg_fingerprint: 4089 D8F2 FDB1 9C98 pkg_vendor: Fedora Project reason: system-python killed by SIGABRT runlevel: N 5 type: CCpp uid: 0
It seems to be fixed by Bug 1451212 - epel.repo: mirrorlist instead of metalink https://bugzilla.redhat.com/show_bug.cgi?id=1451212. The corresponding epel release epel-release-7-10.noarch does fix my same issue.