Re: F27 to F28 Upgrade Fails at Offline Install Time
On 11/7/18 8:47 am, Samuel Sieb wrote: On 07/10/2018 03:17 PM, Stephen Morris wrote: 1). Can anyone shed any light on why the install may have failed and what I need to do to work around it? I'm hoping I don't have to run a 'sudo dnf clean all' and then a 'sudo dnf system-upgrade --refresh --releasever=28 --allowerasing --skip-broken' to download all the packages again. I wonder if the --skip-broken is a problem. Can you solve that before doing the upgrade? I think the packages should still be downloaded. Check in /var/lib/dnf/system-upgrade All the packages were still downloaded. 2). Where can I find the logs that would contain the error message so I can read what it said, given that /var/log/boot.log and /var/log/dnf.log both don't contain the message? Try "dnf system-upgrade log". I issued that command and all it did was tell me there were 6 boots that appeared to have upgrade logs, going right back to the upgrade I did of F23 to F24, but the names it listed were meaningless to me and it didn't provide any path references to where I might find the logs to be able to search them. Following your suggestion I ran a sudo dnf upgrade which produced all the conflict messages that I worked around by adding --allowerasing and --skip-broken, but this time dnf said to try --best and --allowerasing, which I did and that installed all the package updates that caused the conflicts. Having done this I was able to complete the upgrade to F28, but now I have big problems. Under F28 my USB network device doesn't work because dkms can't compile the driver for it, and, either because of this failure or a compile failure the drivers for my mouse haven't been installed by dkms either. Booting into the latest F27 kernel I have installed which is 4.17.3-100, where both drivers have previously successfully compiled and installed (which is how I'm able to send this email), the mouse drivers still compile but the usb device driver fails to compile. Both with the F28 kernel, being 4.17.3-200, and the F27 kernel I mentioned, the compile fails with stdarg.h not found, so I'm now confused about what the F28 upgrade has done. The F28 upgrade should not have updated the F27 kernel source that was already installed and hence there should not be any barriers to compiling the driver against the F27 kernel when under F27 it did compile. regards, Steve ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/BKHGVMEYIAPXUKRLCOOZLZHF6ZN5BOEK/ ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/MBT7SIV74NAZSD6K4IEHNWBOEEEDZFJT/
Re: F27 to F28 Upgrade Fails at Offline Install Time
On 07/11/2018 03:16 PM, Stephen Morris wrote: On 11/7/18 8:47 am, Samuel Sieb wrote: On 07/10/2018 03:17 PM, Stephen Morris wrote: 2). Where can I find the logs that would contain the error message so I can read what it said, given that /var/log/boot.log and /var/log/dnf.log both don't contain the message? Try "dnf system-upgrade log". I issued that command and all it did was tell me there were 6 boots that appeared to have upgrade logs, going right back to the upgrade I did of F23 to F24, but the names it listed were meaningless to me and it didn't provide any path references to where I might find the logs to be able to search them. Each line has a number at the front. Run "dnf system-upgrade log --number=n" to see the corresponding log. Booting into the latest F27 kernel I have installed which is 4.17.3-100, where both drivers have previously successfully compiled and installed (which is how I'm able to send this email), the mouse drivers still compile but the usb device driver fails to compile. Both with the F28 kernel, being 4.17.3-200, and the F27 kernel I mentioned, the compile fails with stdarg.h not found, so I'm now confused about what the F28 upgrade has done. The F28 upgrade should not have updated the F27 kernel source that was already installed and hence there should not be any barriers to compiling the driver against the F27 kernel when under F27 it did compile. Why don't the drivers for the previous kernels still exist? The upgrade shouldn't have removed them. The upgrade will have replaced the -headers package which is what the drivers will compile against. You could try replacing it with an earlier -headers package version, but I doubt that's the problem. This is why I strongly avoid hardware that requires out-of-tree drivers... However, this might be the solution: https://github.com/lwfinger/rtl8723bu/issues/106 ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/EXUSTLVYFZ2GXOPATJWR7DHYU4HZAMLM/
Re: F27 to F28 Upgrade Fails at Offline Install Time
On 07/11/2018 03:16 PM, Stephen Morris wrote: > On 11/7/18 8:47 am, Samuel Sieb wrote: >> On 07/10/2018 03:17 PM, Stephen Morris wrote: >>> 1). Can anyone shed any light on why the install may have failed and >>> what I need to do to work around it? I'm hoping I don't have to run a >>> 'sudo dnf clean all' and then a 'sudo dnf system-upgrade --refresh >>> --releasever=28 --allowerasing --skip-broken' to download all the >>> packages again. >> >> I wonder if the --skip-broken is a problem. Can you solve that before >> doing the upgrade? >> >> I think the packages should still be downloaded. >> Check in /var/lib/dnf/system-upgrade > All the packages were still downloaded. >> >>> 2). Where can I find the logs that would contain the error message so >>> I can read what it said, given that /var/log/boot.log and >>> /var/log/dnf.log both don't contain the message? >> >> Try "dnf system-upgrade log". > > I issued that command and all it did was tell me there were 6 boots that > appeared to have upgrade logs, going right back to the upgrade I did of > F23 to F24, but the names it listed were meaningless to me and it didn't > provide any path references to where I might find the logs to be able to > search them. > > Following your suggestion I ran a sudo dnf upgrade which produced all > the conflict messages that I worked around by adding --allowerasing and > --skip-broken, but this time dnf said to try --best and --allowerasing, > which I did and that installed all the package updates that caused the > conflicts. Having done this I was able to complete the upgrade to F28, > but now I have big problems. > > Under F28 my USB network device doesn't work because dkms can't compile > the driver for it, and, either because of this failure or a compile > failure the drivers for my mouse haven't been installed by dkms either. > > Booting into the latest F27 kernel I have installed which is 4.17.3-100, > where both drivers have previously successfully compiled and installed > (which is how I'm able to send this email), the mouse drivers still > compile but the usb device driver fails to compile. Both with the F28 > kernel, being 4.17.3-200, and the F27 kernel I mentioned, the compile > fails with stdarg.h not found, so I'm now confused about what the F28 > upgrade has done. The F28 upgrade should not have updated the F27 kernel > source that was already installed and hence there should not be any > barriers to compiling the driver against the F27 kernel when under F27 > it did compile. Uhm, did you get the libstdc++-devel RPM installed? It provides the stdarg.h file, specifically: /usr/include/c++/8/tr1/stdarg.h Checking an old F26 VM I have, the equivalent file is: /usr/include/c++/7/tr1/stdarg.h I believe the C compiler will find it if it's installed. Even if it does appear to be installed, a "dnf reinstall libstdc++-devel" should clean up any cruft your messed-up install may have left. -- - Rick Stevens, Systems Engineer, AllDigitalri...@alldigital.com - - AIM/Skype: therps2ICQ: 226437340 Yahoo: origrps2 - -- -- ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/3FCAGQ7JBS6CRCWMYKM5OZZ4A5IYST54/
Re: F27 to F28 Upgrade Fails at Offline Install Time
On Thu, 12 Jul 2018 08:16:06 +1000 Stephen Morris wrote: > Both with the F28 kernel, being 4.17.3-200, and the F27 kernel I > mentioned, the compile fails with stdarg.h not found, so I'm now > confused about what the F28 upgrade has done. The F28 upgrade should > not have updated the F27 kernel source that was already installed and > hence there should not be any barriers to compiling the driver > against the F27 kernel when under F27 it did compile. It sounds like you are either missing the package that provides the C standard library include files, or its location has changed. I doubt it has anything to do with the kernel. On my system, I have the following hits when I use: $ find /usr -iname '*stdarg*' /usr/lib/dietlibc/include/stdarg-cruft.h /usr/lib/dietlibc/include/stdarg.h /usr/lib/gcc/x86_64-redhat-linux/8/include/cross-stdarg.h /usr/lib/gcc/x86_64-redhat-linux/8/include/stdarg.h /usr/lib/gcc/x86_64-redhat-linux/3.4.6/include/stdarg.h /usr/lib64/llvm5.0/lib/clang/5.0.1/include/stdarg.h /usr/lib64/bcc/include/stdarg.h /usr/lib64/clang/6.0.0/include/stdarg.h /usr/lib64/pcc/x86_64-unknown-linux-gnu/1.2.0.DEVEL/include/libpcc_stdarg.h /usr/lib64/pcc/x86_64-unknown-linux-gnu/1.2.0.DEVEL/include/stdarg.h /usr/lib64/perl5/stdarg.ph /usr/share/castxml/clang/include/stdarg.h /usr/share/root/cling/lib/clang/5.0.0/include/stdarg.h /usr/share/splint/imports/stdarg.lcs /usr/share/splint/imports/stdarg.lcl /usr/share/man/man0p/stdarg.h.0p.gz /usr/share/man/man3/stdarg.3.gz /usr/share/sdcc/include/pic16/stdarg.h /usr/share/sdcc/include/stdarg.h /usr/include/efi/efistdarg.h /usr/include/c++/8/tr1/cstdarg /usr/include/c++/8/tr1/stdarg.h /usr/include/c++/8/cstdarg /usr/include/boost/compatibility/cpp_c_headers/cstdarg I think this /usr/lib/gcc/x86_64-redhat-linux/8/include/stdarg.h is the official one, though I'm not sure where the drivers you are trying to compile are looking. You could look at them (use a grep for stdarg) in their source directory. The standard version comes from the gcc package, gcc-8.1.1-1.fc28.x86_64. Is it installed on your system? ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/URYO6MEU5MXGT7OL652CLO3K55VPM5O5/
Re: F27 to F28 Upgrade Fails at Offline Install Time
On 12/7/18 8:38 am, Rick Stevens wrote: On 07/11/2018 03:16 PM, Stephen Morris wrote: On 11/7/18 8:47 am, Samuel Sieb wrote: On 07/10/2018 03:17 PM, Stephen Morris wrote: 1). Can anyone shed any light on why the install may have failed and what I need to do to work around it? I'm hoping I don't have to run a 'sudo dnf clean all' and then a 'sudo dnf system-upgrade --refresh --releasever=28 --allowerasing --skip-broken' to download all the packages again. I wonder if the --skip-broken is a problem. Can you solve that before doing the upgrade? I think the packages should still be downloaded. Check in /var/lib/dnf/system-upgrade All the packages were still downloaded. 2). Where can I find the logs that would contain the error message so I can read what it said, given that /var/log/boot.log and /var/log/dnf.log both don't contain the message? Try "dnf system-upgrade log". I issued that command and all it did was tell me there were 6 boots that appeared to have upgrade logs, going right back to the upgrade I did of F23 to F24, but the names it listed were meaningless to me and it didn't provide any path references to where I might find the logs to be able to search them. Following your suggestion I ran a sudo dnf upgrade which produced all the conflict messages that I worked around by adding --allowerasing and --skip-broken, but this time dnf said to try --best and --allowerasing, which I did and that installed all the package updates that caused the conflicts. Having done this I was able to complete the upgrade to F28, but now I have big problems. Under F28 my USB network device doesn't work because dkms can't compile the driver for it, and, either because of this failure or a compile failure the drivers for my mouse haven't been installed by dkms either. Booting into the latest F27 kernel I have installed which is 4.17.3-100, where both drivers have previously successfully compiled and installed (which is how I'm able to send this email), the mouse drivers still compile but the usb device driver fails to compile. Both with the F28 kernel, being 4.17.3-200, and the F27 kernel I mentioned, the compile fails with stdarg.h not found, so I'm now confused about what the F28 upgrade has done. The F28 upgrade should not have updated the F27 kernel source that was already installed and hence there should not be any barriers to compiling the driver against the F27 kernel when under F27 it did compile. Uhm, did you get the libstdc++-devel RPM installed? It provides the stdarg.h file, specifically: /usr/include/c++/8/tr1/stdarg.h Checking an old F26 VM I have, the equivalent file is: /usr/include/c++/7/tr1/stdarg.h I believe the C compiler will find it if it's installed. Even if it does appear to be installed, a "dnf reinstall libstdc++-devel" should clean up any cruft your messed-up install may have left. I'll check this out, thankyou. The issue is caused because /usr/src/kernels/4.17.3-100.fc27.x86_64/include/linux/kernel.h (this is the location when compiling for the F27 kernel) has the following include statements, and it is the first one that causes the problem: #ifndef _LINUX_KERNEL_H #define _LINUX_KERNEL_H #include #include #include #include #include #include #include #include #include #include #include #include regards, Steve -- - Rick Stevens, Systems Engineer, AllDigitalri...@alldigital.com - - AIM/Skype: therps2ICQ: 226437340 Yahoo: origrps2 - -- -- ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/3FCAGQ7JBS6CRCWMYKM5OZZ4A5IYST54/ ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/CLHKLL7FMVVGRI5LAGTLOVQKWAG35FHJ/
Re: F27 to F28 Upgrade Fails at Offline Install Time
Stephen Morris writes: and that installed all the package updates that caused the conflicts. Having done this I was able to complete the upgrade to F28, but now I have big problems. Under F28 my USB network device doesn't work because dkms can't compile the driver for it, and, either because of this failure or a compile failure the drivers for my mouse haven't been installed by dkms either. Pardon the interruption, but are you saying that you need to use a binary blob for a mouse? A mouse? What kinds of a mouse needs a binary blob driver? pgpWX9LimC6SA.pgp Description: PGP signature ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/K3OF6WBWSYNRXWA2DVP6X375BHTYU5GU/
/etc/motd shows up twice now?
I ssh into my fedora 28 system at work, I get this printed when I login: zooty> ssh -l tweety tomh8022 -- /etc/motd printed here -- Last login: Wed Jul 11 20:21:39 2018 from 127.0.0.1 -- /etc/motd printed yet again here -- tomh> Whyfor is it now printing /etc/motd twice? ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/3DGKOLO3ALBY63LKIT3ICYWBZ5TQMRBP/
Re: /etc/motd shows up twice now?
On Wed, Jul 11, 2018 at 8:37 PM Tom Horsley wrote: > I ssh into my fedora 28 system at work, I get this printed > when I login: > > zooty> ssh -l tweety tomh8022 > -- > /etc/motd printed here > -- > Last login: Wed Jul 11 20:21:39 2018 from 127.0.0.1 > -- > /etc/motd printed yet again here > -- > tomh> > > Whyfor is it now printing /etc/motd twice? > FWIW, you can usually get more information by throwing on the verbose flag, -v a few times, like `ssh -vvv ...` I suspect you have somewhere in your sshd configuration a line like: Banner /etc/motd ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/RQ24TJ7JL6JORXWQLDAFO6GHGQULVCQ6/
Re: /etc/motd shows up twice now?
On 07/12/18 08:27, Tom Horsley wrote: > I ssh into my fedora 28 system at work, I get this printed > when I login: > > zooty> ssh -l tweety tomh8022 > -- > /etc/motd printed here > -- > Last login: Wed Jul 11 20:21:39 2018 from 127.0.0.1 > -- > /etc/motd printed yet again here > -- > tomh> > > Whyfor is it now printing /etc/motd twice? Because there are 2 methods for printing motd. Edit your /etc/ssh/sshd_config and change the PrintMotd from the default "yes" to "no" and/or read the notes in sshd_config.rpmnew which probably now exists on your system. -- Conjecture is just a conclusion based on incomplete information. It isn't a fact. signature.asc Description: OpenPGP digital signature ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/F62JDDQOL5KCCFB6Z3HJHPZH7ESJXUJQ/
Re: /etc/motd shows up twice now?
On 07/12/18 09:45, Ed Greshko wrote: > On 07/12/18 08:27, Tom Horsley wrote: >> I ssh into my fedora 28 system at work, I get this printed >> when I login: >> >> zooty> ssh -l tweety tomh8022 >> -- >> /etc/motd printed here >> -- >> Last login: Wed Jul 11 20:21:39 2018 from 127.0.0.1 >> -- >> /etc/motd printed yet again here >> -- >> tomh> >> >> Whyfor is it now printing /etc/motd twice? > Because there are 2 methods for printing motd. > > Edit your /etc/ssh/sshd_config and change the PrintMotd from the default > "yes" to > "no" and/or read the notes in sshd_config.rpmnew which probably now exists > on your system. Oh, forgot to mention the changelog for openssh-server * Tue Jul 03 2018 Jakub Jelen - 7.7p1-5 + 0.10.3-4 - Disable manual printing of motd by default (#1591381) -- Conjecture is just a conclusion based on incomplete information. It isn't a fact. signature.asc Description: OpenPGP digital signature ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/message/QOPIETIPIBAMLDG6ABMBDMRGODCOC4W6/
