Howdy, I booted to runlevel 3 and followed: https://wiki.mageia.org/en/Mageia_8_Release_Notes#Upgrading_online,_using_urpmi_%28CLI%29 but first, removing dnf and npm, which brought nodejs, none of which do I figure I'm currently using. I had just done 'urpmi --auto-update' and skipped that step. (I ran it after the failed upgrade to check but nothing was updated). I started "script upgrade_log.txt", removemedia'd 7, addmedia'd 8, and urpmi --auto-update --auto --force --download-all --test Initially, many packages were removed due to file conflicts, possibly harmless. For example, from upgrade_log, which I'll attach: The following packages have to be removed for others to be upgraded: gimp-python-2.10.10-1.6.mga7.x86_64 (due to missing pygtk2.0) lib64boost_numpy37_1.68.0-1.68.0-4.mga7.x86_64 (due to missing libpython3.7m.so.1.0()(64bit), due to missing libboost_python37.so.1.68.0()(64bit)) lib64boost_python2-devel-1.68.0-4.mga7.x86_64 (due to unsatisfied boost-devel == 1.68.0-4.mga7) lib64boost_python37_1.68.0-1.68.0-4.mga7.x86_64 (due to missing libpython3.7m.so.1.0()(64bit)) lib64botan2_9-2.9.0-2.1.mga7.x86_64 (due to unsatisfied botan2 == 2.9.0-2.1.mga7) lib64fm_qt6-0.14.1-2.mga7.x86_64 (due to unsatisfied libfm-qt == 0.14.1-2.mga7) lib64kf5kcontacts5-19.04.0-1.mga7.x86_64 (due to unsatisfied kcontacts == 19.04.0-1.mga7) lib64kf5libkdepimakonadi5-19.04.0-1.mga7.x86_64 (due to unsatisfied libkdepim == 2:19.04.0-1.mga7) lib64kpimkdav5-19.04.0-1.mga7.x86_64 ... and installation failed at the end: Installation failed: file /usr/lib64/libmythavcodec.so.58 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x 86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavdevice.so.58 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavfilter.so.7 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavformat.so.58 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 ... Thanks.
The log is 4M, so linked here: https://drive.google.com/uc?export=download&id=1uMnp7HnwhmJ00hvuiT4a63gLMvkAVXFq
The error lines are Installation failed: file /usr/lib64/libmythavcodec.so.58 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavdevice.so.58 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavfilter.so.7 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavformat.so.58 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythavutil.so.56 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythpostproc.so.55 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythswresample.so.3 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/lib64/libmythswscale.so.5 from install of lib64myth31-31.0-20210112.1.mga8.tainted.x86_64 conflicts with file from package lib64myth30-30.0-20200418.1.mga7.tainted.x86_64 file /usr/include/opencv2 from install of opencv-devel-4.5.1-1.mga8.x86_64 conflicts with file from package opencv-devel-3.4.5-2.1.mga7.x86_64 For the opencv-devel package, uninstall it prior to the upgrade and reinstall it if desired after the upgrade. The devel files known to cause problems when upgrading and are usually only installed by Mageia packagers or people compiling a program from a source other then Mageia. Assigning to daviddavid as the person who packaged this version of mythtv
Assignee: bugsquad => geiger.david68210Source RPM: (none) => mythtv-31.0-20210112.1.mga8.taintedCC: (none) => davidwhodgins
Blocks: (none) => 28393
CC: (none) => fri
(In reply to Dave Hodgins from comment #2) > The error lines are... > > For the opencv-devel package, uninstall it prior to the upgrade and reinstall > it if desired after the upgrade. The devel files known to cause problems when > upgrading and are usually only installed by Mageia packagers or people > compiling a program from a source other then Mageia. > > Assigning to daviddavid as the person who packaged this version of mythtv I was reminded of reading about an upgrade issue with devel files and found I was thinking about 32-bit devel files on 64-bit system upgrading MGA6->7: https://wiki.mageia.org/en/Mageia_7_Release_Notes#Upgrading_from_Mageia_6 (I had used the provided command to check for and remove any such at that time to avoid issues going forward.) Other than that, my porous memory does not provide a record of knowing about some general devel-packages problem with distro upgrades. [rolf@x570i ~]$ cat git.txt 20200305-xbmc/ 2020-dvbhdhomerun/ 2020-libhdhomerun/ acestreamplayer/ Adafruit-Pi-Finder/ advancedtomato-arm/ asoundconf/ barry/ bluegriffon-source/ CeroWrtScripts/ devede/ dvbhdhomerun/ etcher/ f3/ f3-6.0/ f3-6.0.zip f3-qt/ gcalcli/ gnash/ gtkam/ h0tw1r3/ hw-probe/ isodumper/ kaccounts-providers/ kaffeine/ leia/ libhdhomerun/ man2html/ memtest86/ mharsch/ namebench/ pan2/ pb-for-desktop/ PhoneInfoga/ ps3mediaserver/ qdirstat/ rtl8821CU/ rtl8821CU-brektrou/ Sachesi/ sane-airscan/ Scripts/ servo/ Spectre-Meltdown-Checker-Automated/ ssr/ tribler/ w_scan2/ xawtv4/ [rolf@x570i ~] This is the contents of my ~/git, where I clone code for most of the packages I've tried to build over recent time. When the build recipe reports a missing dependency, I employ the capable urpm tools and my own frantic hammering IT method to install likely helpful devel packages. As housekeeping is not my forté, many such packages have accumulated and most, if not all (absent kernel-devel) have no immediate or near future utility for my OS. I'm posting from [rolf@x570i ~]$ cat /etc/release Mageia release 8 (Official) for x86_64 [rolf@x570i ~]$ uptime 09:43:32 up 2:43, 3 users, load average: 0.33, 0.33, 0.24 but it took some doing. First, I removed opencv-devel, lib64myth30, and their "parent" programs. I followed the wiki, again, where --test reported "Installation is possible." After --test was removed, circa 5000 packages got updated, all from cache, so fairly quickly, until completion. I rebooted to a black plasma desktop with only a cursor, which responded to the mouse. Eventually, I found konsole from Ctrl-Alt-T and started programs with such as: systemsettings5 & That gave an error of missing libQt5QmlWorkerScript.so.5 and urpmf told me to install lib64qt5qmlworkerscript5 System settings then loaded and a reboot gave me the familiar plasma desktop. Looks good, congrats! I can provide the log or other info, if needed. Thanks!
Title say upgrade mga7>8 fails. I suppose that mean system dont boot to desktop? If so it is severe.
Summary: Runlevel 3 urpmi upgrade MGA7->MGA8 fails on file conflicts => mythtv: upgrade MGA7->MGA8 fails on file conflictsComponent: Release (media or process) => RPM PackagesSeverity: normal => majorPriority: Normal => High
https://wiki.mageia.org/en/Mageia_8_Errata#Various_upgrade_issues
Keywords: (none) => IN_ERRATA8
(In reply to Morgan Leijström from comment #4) > Title say upgrade mga7>8 fails. > I suppose that mean system dont boot to desktop? > If so it is severe. As I described, the --test upgrade failed, did not happen at all. Following the analysis by Dave Hodgins, > > First, I removed opencv-devel, lib64myth30, and their "parent" programs. I > followed the wiki, again, where --test reported "Installation is possible." > After --test was removed, circa 5000 packages got updated, all from cache, > so fairly quickly, until completion. I rebooted to a black plasma desktop > with only a cursor, which responded to the mouse. So, I had interpreted that both opencv* and the myth packages would block any update attempt. After removing, the upgrade completed and booted to a black desktop with cursor. Various reports of such a circumstance with plasma can be found at google. It appears to me a missing lib64qt5qmlworkerscript5 was the cause, in my case, of that symptom since installing it resulted in boot to my normal plasma desktop. I don't see that as related to the topic of this bug, IDK. > > Eventually, I found konsole from Ctrl-Alt-T and started programs with such > as: > systemsettings5 & > That gave an error of missing libQt5QmlWorkerScript.so.5 and urpmf told me > to install lib64qt5qmlworkerscript5 > System settings then loaded and a reboot gave me the familiar plasma desktop. > Looks good, congrats! I can provide the log or other info, if needed. > > Thanks!
Thanks Rolf 1) So a question is why lib64qt5qmlworkerscript5 was missing. 2) for the future, to greatly compress files like this log, issue xz thatfile 3) Someone may suggest removal or altering of the entry in Errata
Created attachment 12417 [details] xz-compressed upgrade log when "successful" Ok, Morgan. I'm attaching the compressed log for the second attempt that got me to MGA8. If anyone needs info to discover the reason for the missing package, let me know. Thanks!
Thanks for the update Rolf. Glad your upgrade has completed ok now. The mythtv package still has to be fixed for the conflicts shown in comment 2 The missing requires for lib64qt5qmlworkerscript5 needs to be analyzed. "urpmq --whatrequires-recursive" shows it is already required by task-plasma5-minimal", and sddm. Rolf, what does "rpm -qa|grep ^task" show?
Rolf, is there anything other then comments in /etc/urpmi/skip.list ?
Ok, the cause of the missing lib64qt5qmlworkerscript5 is now clear to me from the upgrade log. At some time prior to upgrading, task-plasma5-minimal was removed, so the dependencies it needs were not included in the upgrade. Rolf, is there any particular reason task-plasma5-minimal was removed? Anyway, I don't consider that to be a Mageia bug so this bug report is just about the mythtv conflicts.
(In reply to Dave Hodgins from comment #11) > At some time prior to upgrading, task-plasma5-minimal was removed, so the > dependencies it needs were not included in the upgrade. Good. I think we should add in upgrade instructions to check relevant task-* packages are installed; at least for the main desktop. - before starting upgrade. And tell why. It have happened some times users clean out unneeded packages, which cause various problems when there is no task-* package keeping things together.
(In reply to Morgan Leijström from comment #12) > add in upgrade instructions to check relevant task-* > packages are installed Added in https://wiki.mageia.org/en/Mageia_8_Release_Notes#Preparations https://wiki.mageia.org/en/How_to_choose_the_right_Mageia_upgrade_method#Preparations
(In reply to Dave Hodgins from comment #9) > Thanks for the update Rolf. Glad your upgrade has completed ok now. > > The mythtv package still has to be fixed for the conflicts shown in comment 2 > > The missing requires for lib64qt5qmlworkerscript5 needs to be analyzed. > "urpmq --whatrequires-recursive" shows it is already required by > task-plasma5-minimal", and sddm. [rolf@x570i ~]$ rpm -q task-plasma5-minimal sddm package task-plasma5-minimal is not installed sddm-0.19.0-14.mga8 [rolf@x570i ~]$ I use sddm and might wonder why that wouldn't pull in lib64qt5qmlworkerscript5 > > Rolf, what does "rpm -qa|grep ^task" show? [rolf@x570i ~]$ rpm -qa|grep ^task task-xfce-4.14-2.mga8 task-lxqt-minimal-0.16.0-1.mga8 task-printing-server-2018-6.mga8 task-obsolete-8-294.mga8 task-printing-hp-2018-6.mga8 task-xfce-minimal-4.14-2.mga8 task-codec-audio-6-4.mga8.tainted task-gnome-3.38.0-1.mga8 task-edu-it-1-19.mga8 task-cinnamon-minimal-4.8-2.mga8 task-obsolete-nonfree-8-4.mga8.nonfree task-cinnamon-4.8-2.mga8 task-codec-video-6-4.mga8.tainted task-x11-1-10.mga8 task-c++-devel-2011.0-9.mga8 task-gnome-minimal-3.38.0-1.mga8 task-edu-1-19.mga8 task-python3-1-4.mga8 task-xapps-1.8-4.mga8 task-lxqt-0.16.0-1.mga8 task-c-devel-2011.0-9.mga8 task-pulseaudio-2018.0-2.mga8 [rolf@x570i ~]$ (In reply to Dave Hodgins from comment #10) > Rolf, is there anything other then comments in /etc/urpmi/skip.list ? [rolf@x570i ~]$ sudo cat /etc/urpmi/skip.list # Here you can specify the packages that won't be upgraded automatically # for example, to exclude all apache packages : # /^apache/ youtube-dl #/^timezone/ #/^kernel/ [rolf@x570i ~]$ (In reply to Dave Hodgins from comment #11) > Ok, the cause of the missing lib64qt5qmlworkerscript5 is now clear to me from > the upgrade log. > > At some time prior to upgrading, task-plasma5-minimal was removed, so the > dependencies it needs were not included in the upgrade. > > Rolf, is there any particular reason task-plasma5-minimal was removed? I have a hunch, fwiw: [root@x570i rolf]# history | grep orphans 464 urpme --auto-orphans 1001 history | grep orphans [root@x570i rolf]# I have almost always feared --auto-orphans and avoided it. Perhaps, I had some inkling it was working better? IDK. I don't think we have evidence that I actually went through with it, in case the command asks for confirmation, as I would expect. > Anyway, I don't consider that to be a Mageia bug so this bug report is just > about > the mythtv conflicts. Isn't > file /usr/include/opencv2 from install of opencv-devel-4.5.1-1.mga8.x86_64 > conflicts with file from package opencv-devel-3.4.5-2.1.mga7.x86_64 a problem, also? Thanks!
I think I'm wrong about the cause of the problem. sddm did pull in qtdeclarative5-5.15.2-1.mga8.x86_64.rpm $ rpm -q --requires qtdeclarative5|grep -i qt5qmlworker libQt5QmlWorkerScript.so.5()(64bit) libQt5QmlWorkerScript.so.5(Qt_5)(64bit) libQt5QmlWorkerScript.so.5(Qt_5_PRIVATE_API)(64bit) $ urpmq --whatprovides 'libQt5QmlWorkerScript.so.5()(64bit)' lib64qt5qmlworkerscript5 Was qtdeclarative5 actually installed? If not, then when the transaction to install mythtv failed, all of the packages in the same transaction failed too. I'm getting confused. :-)
(In reply to Rolf Pedersen from comment #14) > Isn't > > file /usr/include/opencv2 from install of opencv-devel-4.5.1-1.mga8.x86_64 > > conflicts with file from package opencv-devel-3.4.5-2.1.mga7.x86_64 > a problem, also? Yes. I was mixing up the fact that 32 and 64 bit versions of a devel package cannot be installed in a system. The problem there is that the directory /usr/include/opencv2 is being replaced by a symlink /usr/include/opencv2 -> opencv4/opencv2/ So the package opencv-devel needs to be fixed too.
For mythtv, similarly - On Mageia 7 # ll /usr/lib64/libmythavcodec.so* lrwxrwxrwx 1 root root 27 May 6 2020 /usr/lib64/libmythavcodec.so.58 -> libmythavcodec.so.58.18.100* -rwxr-xr-x 1 root root 13275600 May 6 2020 /usr/lib64/libmythavcodec.so.58.18.100* On Mageia 8 # ll /usr/lib64/libmythavcodec.so* lrwxrwxrwx 1 root root 27 Jan 12 09:15 /usr/lib64/libmythavcodec.so.58 -> libmythavcodec.so.58.54.100* -rwxr-xr-x 1 root root 13843120 Jan 12 09:18 /usr/lib64/libmythavcodec.so.58.54.100* So it's trying to replace a symlink pointing to one file with a symlink pointing to a different file. I would have thought that would be ok, but apparently not. I haven't checked but expect the rest of the mythtv file conflicts to be the same.
Joseph, see comment 16 about opencv file conflict on m7 to m8 upgrade.
CC: (none) => joequant
(In reply to Dave Hodgins from comment #15) > I think I'm wrong about the cause of the problem. > > sddm did pull in qtdeclarative5-5.15.2-1.mga8.x86_64.rpm > $ rpm -q --requires qtdeclarative5|grep -i qt5qmlworker > libQt5QmlWorkerScript.so.5()(64bit) > libQt5QmlWorkerScript.so.5(Qt_5)(64bit) > libQt5QmlWorkerScript.so.5(Qt_5_PRIVATE_API)(64bit) > $ urpmq --whatprovides 'libQt5QmlWorkerScript.so.5()(64bit)' > lib64qt5qmlworkerscript5 > > Was qtdeclarative5 actually installed? My initial boot of Mageia 8 seems apropos... (In reply to Rolf Pedersen from comment #3) > (In reply to Dave Hodgins from comment #2) ... > I rebooted to a black plasma desktop > with only a cursor, which responded to the mouse. > > Eventually, I found konsole from Ctrl-Alt-T and started programs with such > as: > systemsettings5 & > That gave an error of missing libQt5QmlWorkerScript.so.5 and urpmf told me > to install lib64qt5qmlworkerscript5 > System settings then loaded and a reboot gave me the familiar plasma desktop. ... I have omitted a detail. The error that led me to install lib64qt5qmlworkerscript5 began with a reference to qtdeclarative5 and the following shows what I went through, mistakenly thinking the problem was with my (installed) qtdeclarative5. I think, now, the error was saying qtdeclarative5 looked for libQt5QmlWorkerScript.so.5 but could not find it. When urpmf showed that (not installed) lib64qt5qmlworkerscript5 provided that file, I installed it and the rest is history ;) [rolf@x570i 8]$ history | grep qtdeclarative5 919 sudo urpmi qtdeclarative5 920 sudo urpmi qtdeclarative5 --replacefiles 921 sudo urpmi qtdeclarative5 --replacefiles --force 922 sudo urpmi qtdeclarative5 --replacefiles --allow-force 923 sudo urpmi qtdeclarative5 --replacefiles --replacefiles 924 urpmq --sources qtdeclarative5 925 rpm -Uvh --replacefiles --replacepkgs http://mirrors.kernel.org/mageia/distrib/8/x86_64/media/core/release/qtdeclarative5-5.15.2-1.mga8.x86_64.rpm 926 sudo rpm -Uvh --replacefiles --replacepkgs http://mirrors.kernel.org/mageia/distrib/8/x86_64/media/core/release/qtdeclarative5-5.15.2-1.mga8.x86_64.rpm 1004 history | grep qtdeclarative5 [rolf@x570i 8]$ In short, qtdeclarative5 was installed, lib64qt5qmlworkerscript5 was not. Thanks! > If not, then when the transaction > to install mythtv failed, all of the packages in the same transaction failed > too. > > I'm getting confused. :-)
Ping David GEIGER for mythtv file conflicts and Joseph Wang for opencv file conflicts. These must be fixed before upgrading using mgaapplet can be enabled.
Should be fixed with mythtv-31.0-20210112.1.1.mga8 in Core/Updates_testing and Tainted/Updates_testing repo!
Confirmed mythtv fixed. Installed all packages starting with mythtv in an m7 snapshot, removed the m7 repos, added the m8 repos, enabled core updates testing, and ran the urpmi command to install all of the mythtv packages again. Had to remove mythtv-plugin-gallery from the command as that package does not exist in Mageia 8. Install of the mythtv packages along with all of it's dependencies went smoothly with no file conflicts. Will open a new bug report for opencv file conflict and then prepare this one for use to push the m8 mythtv update.
I've opened bug 28522 for opencv. Advisory committed to svn. Validating the update.
Whiteboard: (none) => MGA8-64-OKKeywords: (none) => advisory, validated_update
(In reply to Dave Hodgins from comment #22) > Had to > remove mythtv-plugin-gallery from the command as that package does not exist > in Mageia 8. I assume that during update using our recommended procedures this will be taken care of automatically?
An update for this issue has been pushed to the Mageia Updates repository. https://advisories.mageia.org/MGAA-2021-0037.html
Status: NEW => RESOLVEDResolution: (none) => FIXED
Marked as fixed in Errata
(In reply to Morgan Leijström from comment #24) > (In reply to Dave Hodgins from comment #22) > > Had to > > remove mythtv-plugin-gallery from the command as that package does not exist > > in Mageia 8. > > I assume that during update using our recommended procedures this will be > taken care of automatically? I only removed it from my urpmi command. I didn't remove it from the system prior to testing the upgraded package would install ok. I didn't check to see if it was left as an orphan or removed, and reverted the testing snapshot test after ensuring the conflicts were fixed. Checking it's requires in Mageia 7, it must have been removed. I do not have the hardware needed to test that mythtv actually works.