Requesting Suggestions: System broken after 7.6 Update

General support questions
tonybKDE
Posts: 11
Joined: 2018/12/06 01:52:48

Requesting Suggestions: System broken after 7.6 Update

Post by tonybKDE » 2018/12/08 16:24:46

This is a long-ish post asking for help fixing issues that occurred with the CentOS 7.6 (1810) update.
Apologies for the length but I’m trying to give enough information to allow someone to give me some suggestions for what to do next.

I made some errors in updating a CentOS machine to 7.6 and as a result it’s now is very marginally functional. It’s an older Lenovo D30 workstation, very powerful with lots of memory and CPU but with a vintage 2012 NVidia Quadro 2000 card. I am running the NVidia proprietary driver. Within the last three months or so support for my card has lapsed and YUM’s nvidia plug-in has correctly stopped updating the driver and I am still running the 390 legacy version.

My errors were (I think):
  • I was running KDE at the time ofthe upgrade
    I was running NVidia at the time of the upgrade
    I ignored some problems with xorg* packages and excluded them from the update
    I ignored some issues with the centos-sclo repo and excluded it from the update
As best I can reconstruct from my yum history here is the command I ran:

Code: Select all

[tt]Command Line   : update --exclude=filesystem,xorg* --disablerepo=centos-sclo-sclo[/tt]
900+ packages were updated and several, including some GNOME packages, were obsoleted. The only unusual messages I recorded were related to configuration of the “infinality” font package.

Symptoms are
GNOME will no longer start. KDE, xfce, and IceWM will start.

The NVidia driver still functions but nvidia-settings configuration GUI doesn’t work

Code: Select all

[tt]ERROR: /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates
       /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates
       /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates
       /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates


ERROR: A problem occured when loading the GUI library. Please check your installation and library path. You may need to specify this library when
       calling nvidia-settings. Please run `nvidia-settings --help` for usage information[/tt].

At least one program, the shutter screensaver, doesn’t work

Code: Select all

[tt]$ shutter
WARNING: Gtk2::AppIndicator is missing --> there will be no icon showing up in the status bar when running Unity!

Fontconfig error: failed reading config file
Fontconfig error: Cannot load config file "infinality/conf.d"


Fontconfig error: Cannot load config file "infinality/conf.d"                                                                                          
GdkPixbuf-LOG **: gdk_pixbuf_from_pixdata() called on: at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.          
GdkPixbuf-LOG **:       Encoding raw at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.                            
GdkPixbuf-LOG **:       Dimensions: 16 x 16 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Rowstride: 64, Length: 1048 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Copy pixels == false at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **: gdk_pixbuf_from_pixdata() called on: at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Encoding raw at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Dimensions: 16 x 16 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Rowstride: 64, Length: 1048 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Copy pixels == false at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.[/tt]
The nedit editor appears to be affected by the infinality font issue. It starts from the command line and runs, but returns the following to the console:

Code: Select all

[tt]Fontconfig error: failed reading config file
Fontconfig error: Cannot load config file "infinality/conf.d"[/tt]
The xorg issues that caused me to exclude the updates persist. Several xorg packages show up in a yum check-update, but trying to actually update them produces the following:

Code: Select all

[tt]sudo yum update xorg*
Loaded plugins: fastestmirror, langpacks, nvidia
Loading mirror speeds from cached hostfile
 * base: distro.ibiblio.org
 * elrepo: dfw.mirror.rackspace.com
 * elrepo-extras: dfw.mirror.rackspace.com
 * elrepo-kernel: dfw.mirror.rackspace.com
 * epel: archive.linux.duke.edu
 * epel-testing: archive.linux.duke.edu
 * extras: mirror.atlanticmetro.net
 * nux-dextop: li.nux.ro
 * updates: centos.mirror.lstn.net
[nvidia]: excluding kmod-nvidia-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.73-2.el7_6.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.78-2.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-libs-410.78-2.el7_6.elrepo.i686
[nvidia]: excluding nvidia-x11-drv-libs-410.78-2.el7_6.elrepo.x86_64
Resolving Dependencies
--> Running transaction check
---> Package xorg-x11-drv-ati.x86_64 0:7.10.0-1.el7 will be updated
---> Package xorg-x11-drv-ati.x86_64 0:18.0.1-1.el7 will be an update
---> Package xorg-x11-drv-fbdev.x86_64 0:0.4.3-25.el7 will be updated
---> Package xorg-x11-drv-fbdev.x86_64 0:0.5.0-1.el7 will be an update
---> Package xorg-x11-drv-intel.x86_64 0:2.99.917-27.20160929.el7 will be updated
---> Package xorg-x11-drv-intel.x86_64 0:2.99.917-28.20180530.el7 will be an update
---> Package xorg-x11-drv-nouveau.x86_64 1:1.0.13-3.el7 will be updated
---> Package xorg-x11-drv-nouveau.x86_64 1:1.0.15-1.el7 will be an update
---> Package xorg-x11-drv-qxl.x86_64 0:0.1.5-3.el7 will be updated
---> Package xorg-x11-drv-qxl.x86_64 0:0.1.5-4.el7.1 will be an update
---> Package xorg-x11-drv-v4l.x86_64 0:0.2.0-47.el7 will be updated
---> Package xorg-x11-drv-v4l.x86_64 0:0.2.0-49.el7 will be an update
---> Package xorg-x11-drv-vesa.x86_64 0:2.3.2-25.1.el7 will be updated
---> Package xorg-x11-drv-vesa.x86_64 0:2.4.0-1.el7 will be an update
---> Package xorg-x11-drv-vmware.x86_64 0:13.2.1-1.el7 will be updated
---> Package xorg-x11-drv-vmware.x86_64 0:13.2.1-1.el7.1 will be an update
---> Package xorg-x11-server-Xorg.x86_64 0:1.19.5-5.1.el7_5.0.1 will be updated
--> Processing Dependency: xserver-abi(videodrv-23) >= 0 for package: xorg-x11-drv-dummy-0.3.8-1.xpra1.el7_4.x86_64
---> Package xorg-x11-server-Xorg.x86_64 0:1.20.1-5.1.el7 will be an update
--> Finished Dependency Resolution
Error: Package: xorg-x11-drv-dummy-0.3.8-1.xpra1.el7_4.x86_64 (@el7-local)
           Requires: xserver-abi(videodrv-23) >= 0
           Removing: xorg-x11-server-Xorg-1.19.5-5.1.el7_5.0.1.x86_64 (@updates)
               xserver-abi(videodrv-23) = 0
           Updated By: xorg-x11-server-Xorg-1.20.1-5.1.el7.x86_64 (c7-updates)
              ~xserver-abi(videodrv-24) = 0
 You could try using --skip-broken to work around the problem
 You could try running: rpm -Va --nofiles --nodigest[/tt]
If I could do so I would back out this entire update and start over. But with the updates to the kernel, the various kernel modules, and the version change to 7.6 (resulting in back-level versions of many packages being no longer available) I don’t think this is an option.

What’s the best way to approach this problem?

User avatar
TrevorH
Forum Moderator
Posts: 24569
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: Requesting Suggestions: System broken after 7.6 Update

Post by TrevorH » 2018/12/08 18:13:12

FT_Get_Var_Design_Coordinates
Run qpm -qa \*infinality\* and see what packages match that. If those are from the nux-dextop repo then I think you either have to ask him to update those packages (or maybe there are rebuilt versions already in his -testing repo?) or remove them as they're not compatible with the 7.6 version of freetype. That will also fix half the other problems in your post.

I suspect you need newer kmod-nvidia packages but since your cards have been retired, you probably need to switch from the main kmod-nvidia packages to one of the legacy ones, most likely kmod-nvidia390*. That should also let you update the xorg packages too.
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

desertcat
Posts: 518
Joined: 2014/08/07 02:17:29
Location: Tucson, AZ

Re: Requesting Suggestions: System broken after 7.6 Update

Post by desertcat » 2018/12/09 10:49:17

tonybKDE wrote:
2018/12/08 16:24:46
This is a long-ish post asking for help fixing issues that occurred with the CentOS 7.6 (1810) update.
Apologies for the length but I’m trying to give enough information to allow someone to give me some suggestions for what to do next.

I made some errors in updating a CentOS machine to 7.6 and as a result it’s now is very marginally functional. It’s an older Lenovo D30 workstation, very powerful with lots of memory and CPU but with a vintage 2012 NVidia Quadro 2000 card. I am running the NVidia proprietary driver. Within the last three months or so support for my card has lapsed and YUM’s nvidia plug-in has correctly stopped updating the driver and I am still running the 390 legacy version.

My errors were (I think):
  • I was running KDE at the time ofthe upgrade
    I was running NVidia at the time of the upgrade
    I ignored some problems with xorg* packages and excluded them from the update
    I ignored some issues with the centos-sclo repo and excluded it from the update
As best I can reconstruct from my yum history here is the command I ran:

Code: Select all

[tt]Command Line   : update --exclude=filesystem,xorg* --disablerepo=centos-sclo-sclo[/tt]
900+ packages were updated and several, including some GNOME packages, were obsoleted. The only unusual messages I recorded were related to configuration of the “infinality” font package.

Symptoms are
GNOME will no longer start. KDE, xfce, and IceWM will start.

The NVidia driver still functions but nvidia-settings configuration GUI doesn’t work

Code: Select all

[tt]ERROR: /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates
       /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates
       /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates
       /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinates


ERROR: A problem occured when loading the GUI library. Please check your installation and library path. You may need to specify this library when
       calling nvidia-settings. Please run `nvidia-settings --help` for usage information[/tt].

At least one program, the shutter screensaver, doesn’t work

Code: Select all

[tt]$ shutter
WARNING: Gtk2::AppIndicator is missing --> there will be no icon showing up in the status bar when running Unity!

Fontconfig error: failed reading config file
Fontconfig error: Cannot load config file "infinality/conf.d"


Fontconfig error: Cannot load config file "infinality/conf.d"                                                                                          
GdkPixbuf-LOG **: gdk_pixbuf_from_pixdata() called on: at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.          
GdkPixbuf-LOG **:       Encoding raw at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.                            
GdkPixbuf-LOG **:       Dimensions: 16 x 16 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Rowstride: 64, Length: 1048 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Copy pixels == false at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **: gdk_pixbuf_from_pixdata() called on: at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Encoding raw at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Dimensions: 16 x 16 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Rowstride: 64, Length: 1048 at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.
GdkPixbuf-LOG **:       Copy pixels == false at /usr/share/perl5/vendor_perl/Shutter/App/SimpleDialogs.pm line 254, <DATA> line 19.[/tt]
The nedit editor appears to be affected by the infinality font issue. It starts from the command line and runs, but returns the following to the console:

Code: Select all

[tt]Fontconfig error: failed reading config file
Fontconfig error: Cannot load config file "infinality/conf.d"[/tt]
The xorg issues that caused me to exclude the updates persist. Several xorg packages show up in a yum check-update, but trying to actually update them produces the following:

Code: Select all

[tt]sudo yum update xorg*
Loaded plugins: fastestmirror, langpacks, nvidia
Loading mirror speeds from cached hostfile
 * base: distro.ibiblio.org
 * elrepo: dfw.mirror.rackspace.com
 * elrepo-extras: dfw.mirror.rackspace.com
 * elrepo-kernel: dfw.mirror.rackspace.com
 * epel: archive.linux.duke.edu
 * epel-testing: archive.linux.duke.edu
 * extras: mirror.atlanticmetro.net
 * nux-dextop: li.nux.ro
 * updates: centos.mirror.lstn.net
[nvidia]: excluding kmod-nvidia-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.73-2.el7_6.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.78-2.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-libs-410.78-2.el7_6.elrepo.i686
[nvidia]: excluding nvidia-x11-drv-libs-410.78-2.el7_6.elrepo.x86_64
Resolving Dependencies
--> Running transaction check
---> Package xorg-x11-drv-ati.x86_64 0:7.10.0-1.el7 will be updated
---> Package xorg-x11-drv-ati.x86_64 0:18.0.1-1.el7 will be an update
---> Package xorg-x11-drv-fbdev.x86_64 0:0.4.3-25.el7 will be updated
---> Package xorg-x11-drv-fbdev.x86_64 0:0.5.0-1.el7 will be an update
---> Package xorg-x11-drv-intel.x86_64 0:2.99.917-27.20160929.el7 will be updated
---> Package xorg-x11-drv-intel.x86_64 0:2.99.917-28.20180530.el7 will be an update
---> Package xorg-x11-drv-nouveau.x86_64 1:1.0.13-3.el7 will be updated
---> Package xorg-x11-drv-nouveau.x86_64 1:1.0.15-1.el7 will be an update
---> Package xorg-x11-drv-qxl.x86_64 0:0.1.5-3.el7 will be updated
---> Package xorg-x11-drv-qxl.x86_64 0:0.1.5-4.el7.1 will be an update
---> Package xorg-x11-drv-v4l.x86_64 0:0.2.0-47.el7 will be updated
---> Package xorg-x11-drv-v4l.x86_64 0:0.2.0-49.el7 will be an update
---> Package xorg-x11-drv-vesa.x86_64 0:2.3.2-25.1.el7 will be updated
---> Package xorg-x11-drv-vesa.x86_64 0:2.4.0-1.el7 will be an update
---> Package xorg-x11-drv-vmware.x86_64 0:13.2.1-1.el7 will be updated
---> Package xorg-x11-drv-vmware.x86_64 0:13.2.1-1.el7.1 will be an update
---> Package xorg-x11-server-Xorg.x86_64 0:1.19.5-5.1.el7_5.0.1 will be updated
--> Processing Dependency: xserver-abi(videodrv-23) >= 0 for package: xorg-x11-drv-dummy-0.3.8-1.xpra1.el7_4.x86_64
---> Package xorg-x11-server-Xorg.x86_64 0:1.20.1-5.1.el7 will be an update
--> Finished Dependency Resolution
Error: Package: xorg-x11-drv-dummy-0.3.8-1.xpra1.el7_4.x86_64 (@el7-local)
           Requires: xserver-abi(videodrv-23) >= 0
           Removing: xorg-x11-server-Xorg-1.19.5-5.1.el7_5.0.1.x86_64 (@updates)
               xserver-abi(videodrv-23) = 0
           Updated By: xorg-x11-server-Xorg-1.20.1-5.1.el7.x86_64 (c7-updates)
              ~xserver-abi(videodrv-24) = 0
 You could try using --skip-broken to work around the problem
 You could try running: rpm -Va --nofiles --nodigest[/tt]
If I could do so I would back out this entire update and start over. But with the updates to the kernel, the various kernel modules, and the version change to 7.6 (resulting in back-level versions of many packages being no longer available) I don’t think this is an option.

What’s the best way to approach this problem?
OK is this too dumb a time to ask if you have BACKED UP YOUR DRIVE?!? Both 7.5. and 7.6 have not been too kind to me, but I had BACKED UP MY Drive and even though I toasted my machine I was able to restore 95+% (or at least as far back as 10 Nov) pus I have specific directories that are were less than a week old.

IF you have a BACKUP simply roll the machine back. Simple as that.

OK what IF you DID NOT BACKUP your machine? The odds go way up that you are partially if not totally hosed. IF you still have access to a konsole try doing this: as su run the command yum group list. That will tell you what groups are available to you. From there run yum group install "[name of group]" One of the group that should be listed is GNOME Desktop. If so run yum group install "GNOME Desktop". If you are lucky it might update all the Packages and Dependencies needed by GNOME. If so login as either GNOME or GNOME Classic and be on your merry way.

OK what if you get some message saying in effect that all packages are the most most recent. "Nothing to do."

Here is where you need to make a decision -- run as KDE (or MATE, Xfce, etc.). You might as well go for broke and simply run yum update. But before you can do that you need to deal with that ERROR message:

Error: Package: xorg-x11-drv-dummy-0.3.8-1.xpra1.el7_4.x86_64 (@el7-local). It *may* be simply safe to remove this package as it refers to CentOS 7.4 [ xorg-x11-drv-dummy-0.3.8-1.xpra1.el7_4.x86_64, then unless there is some mad reason simply run yum update and remove the "--exclude=..." options. With LUCK it will then go out and fix all the broken parts, including GNOME.

If after this you are probably 95% hosed. First BACKUP your /home directory. Second download a copy of the CentOS-7.6-1810 DVD.iso burn it to a CD, check the checksum to make sure you got a clean download and burn. Then simply do a fresh re-install. Giant PITA. Just did it myself after I sincerely toasted my test machine. I toasted my workstation TWICE, first by overconfidence and missed something, the second time on purpose to eliminate any source of conflict. BUT... I HAD A BACKUP which I rolled the machine back to. Even as we speak I am still sitting at 7.5. If worse comes to worse before the end of the year I'll simply update the machine and deal with the BUG, as I discovered a workaround -- it still would be a Giant PITA, but I could live with it if need be. Doing a complete reinstall would be a major bummer, but that might be your only option. Just make sure you backup your entire /home directory.

For your sake I hope you backup the machine. It might take a couple hours to roll it back, compared to the other options which could take DAYS!! A few hours is not such a bad thing.

Wishing you a Happy Holiday Season, and a Safe and Prosperous New Year.

tonybKDE
Posts: 11
Joined: 2018/12/06 01:52:48

Re: Requesting Suggestions: System broken after 7.6 Update

Post by tonybKDE » 2018/12/10 18:33:20

Many thanks TrevorH and desertcat for the suggestions.

Here are a couple updates. I determined that there are only two \*infinality\* packages. They depend on one another, but nothing else appears to need them (used rpm --whatrequires to determine this. So I removed them both. Now indeed nedit starts up without the previously-reported fontconfig errors. Progress!

Next try rpm -q --whatrequires xorg-x11-drv-dummy.x86_64. Nothing depends on it, so sudo yum remove xorg-x11-drv-dummy.x86_64 gets rid of it, no problem.

Finally (still in the same session)
cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module 390.42 Sat Mar 3 04:10:22 PST 2018
GCC version: gcc version 4.8.5 20150623 (Red Hat 4.8.5-36) (GCC)


Looks like I do have the 390.xx kernel module. Hm. Try nvidia-settings. Works! I think the correct kernel module comes as a result of having dkms-nvidia installed.

Ran a yum update. All packages (except filesystem, which has caused problems for me for a looooong time) updated properly.

Time for a reboot. Uh-oh, now login screen doesn't appear. Can't startx from a terminal. Need to revisit grub startup line parameters which I modified about a year or so ago to get nvidia to start properly.

Will post an update when I have it.

desertcat, my backup strategy is not great (nonexistent). This is a test machine, though, and my "production" work machine is still going. I do back up the /home on that machine, but in a circumstance like this one where 900+ packages were updated it seems unlikely that any backup short of an entire disk image would not really help. Indeed a re-install might be in the cards. But thank you very much for your comments.

desertcat
Posts: 518
Joined: 2014/08/07 02:17:29
Location: Tucson, AZ

Re: Requesting Suggestions: System broken after 7.6 Update

Post by desertcat » 2018/12/11 20:49:39

tonybKDE wrote:
2018/12/10 18:33:20

desertcat, my backup strategy is not great (nonexistent). This is a test machine, though, and my "production" work machine is still going. I do back up the /home on that machine, but in a circumstance like this one where 900+ packages were updated it seems unlikely that any backup short of an entire disk image would not really help. Indeed a re-install might be in the cards. But thank you very much for your comments.
Me?!? I usually do a full -- entire disk -- backup both right before a new release -- just in case the roll over blows up the machine -- and right after I have finally got the new install rocking and rolling. I use the dd command for this purpose. dd if=/dev/drive-to-be-backed-up conv=sync,noerror bs=64K | gzip -c > CentOS-Major_Release.x-year-and-month_machine-name_date-of-backup.dd.gz. That's what saved me this time. I use backupninja and set it up for weekly backups, of all the key partitions. Just in case something gets toasted in a particular partition or file, I can restore it.

I am still in the throws of re-configuring my test bed after doing a full re-install of 7.6. It lies outside of my network and is not on my LAN thus not backuped, but every machine on the LAN is backuped. My workstation was rolled back to 7.5 to figure out a problem I was having, I have a workaround, clumsy and a bit of a PITA, but it will work. Once I have both my workstation and test machine fully up and running -- probably by the end of the year -- I'll do a full backup the workstation, and then I'll seriously look at at GNOME and start experimenting with it on my test machine.

Have a Safe Holiday Season, and a Happy and Prosperous New Year.

tonybKDE
Posts: 11
Joined: 2018/12/06 01:52:48

Re: Requesting Suggestions: System broken after 7.6 Update

Post by tonybKDE » 2018/12/12 14:22:18

desertcat, there's only one thing to say in a situation like this: you're right. 100%. I was entirely too cavalier with this update, probably spoiled by years of successful (good luck, basically) upgrades with both Red Hat Enterprise and CentOS.

But here we are. I'm fortunate that I have a test machine to experiment with.

Recall from my earlier posts that I had some success by removing the infinality packages and the one xorg package from CentOS 7.4. But on the next boot, the GUI login fails to appear and I'm limited to logging in to a text console. This is what I now see in /var/log/Xorg.0.log

Code: Select all

[    25.100] ================ WARNING WARNING WARNING WARNING ================
[    25.100] This server has a video driver ABI version of 24.0 that this
driver does not officially support.  Please check
http://www.nvidia.com/ for driver updates or downgrade to an X
server with a supported driver ABI.
[    25.100] =================================================================
[    25.100] (EE) NVIDIA: Use the -ignoreABI option to override this check.
[    25.100] (II) UnloadModule: "nvidia"
[    25.100] (II) Unloading nvidia
[    25.100] (EE) Failed to load module "nvidia" (unknown error, 0)
[    25.100] (II) LoadModule: "nouveau"
[    25.100] (II) Loading /usr/lib64/xorg/modules/drivers/nouveau_drv.so
It appears to load nouveau successfully but I guess it's probably conflicting in some way with either my boot parameters or remnants of NVidia that are around somehow. So my plan is to just see if I can get the GUI running with nouveau. First I will try not uninstalling NVidia, and if that doesn't work I'll remove *nvidia*

Right now my /etc/default/grub has this in it; pretty clear some change is needed :

Code: Select all

GRUB_CMDLINE_LINUX="nouveau.modeset=0 rd.driver.blacklist=nouveau rd.driver.blacklist=nouveau rdblacklist=nouveau crashkernel=auto rd
.luks.uuid=luks-eb72f7d2-b4a6-4660-9794-ac54ba62de32 rd.lvm.lv=privg/lv_root rd.lvm.lv=privg/lv_swap rd.lvm.lv=privg/lv_usr nomodeset
"
My plan is as follows:
  • Edit boot parameters to remove nouveau.modeset=0 and change rd.driver.blacklist=nouveau to rd.driver.blacklist=nvidia
    Comment out nouveau blacklisting in /etc/modprobe.d/nouveau_blacklist.conf or /usr/lib/modprobe.d/nvidia.conf (I think I have the latter)
    Create the file /etc/X11/xorg.conf.d/20-nouveau.conf with the following content:

Code: Select all

Section "Device"
    Identifier "Nvidia card"
    Driver "nouveau"
EndSection
I believe this will allow nouveau to start while NVidia is still installed, but if I can't get it going I'll remove the NVidia packages.

Trevor has suggested adding rhgb to the boot command line. I'll certainly try that but in the past that has been implicated in problems I had getting the LUKS encryption prompt to show up early in the boot process. My "normal" operation has this appearing on the console in text mode during the boot process, where I can type it in. The login GUI shows up much later at the end of the process.

I noticed another issue today when trying to mount network drives. Something is wrong with Samba/CIFS. Not sure what. Hopefully will be able to debug that a little more once I get my desktop back.

User avatar
TrevorH
Forum Moderator
Posts: 24569
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: Requesting Suggestions: System broken after 7.6 Update

Post by TrevorH » 2018/12/12 16:14:07

No, I suggested removing rhgb quiet if they are present. They hide useful information. However since you've posted the xorg log file which shows that you have the old nvidia driver installed, where are you getting your nvidia driver from ? If it's ELRepo then it looks like you're missing an update. If it's not ELRepo then why not?
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

tonybKDE
Posts: 11
Joined: 2018/12/06 01:52:48

Re: Requesting Suggestions: System broken after 7.6 Update

Post by tonybKDE » 2018/12/12 16:50:05

Thank you TrevorH, sorry for mis-reading your advice on rhbg setting. It's not in my boot line so I should be seeing everything.

This machine is in a corporate environment and nvidia comes from a local repo set up by our administrator, as far as I can tell. I am not really familiar with the details of the yum display but it is highlighting several packages. I'll study up on what this means but if you have comments/suggestions in the meantime they will surely be appreciated.

Note that the nvidia plugin for yum is preventing updates beyond the 390 legacy driver due to support for my Quadro 2000 NVidia card having expired a month or two ago with the release of the latest driver.

Loaded plugins: fastestmirror, langpacks, nvidia
Loading mirror speeds from cached hostfile
* base: mirror.atlanticmetro.net
* elrepo: dfw.mirror.rackspace.com
* elrepo-extras: dfw.mirror.rackspace.com
* elrepo-kernel: dfw.mirror.rackspace.com
* epel: archive.linux.duke.edu
* epel-testing: archive.linux.duke.edu
* extras: ewr.edge.kernel.org
* nux-dextop: li.nux.ro
* updates: mirror.atlantic.net
[nvidia]: excluding kmod-nvidia-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.73-2.el7_6.elrepo.x86_64
[nvidia]: excluding kmod-nvidia-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-410.78-2.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.66-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.73-1.el7_5.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-32bit-410.78-1.el7_6.elrepo.x86_64
[nvidia]: excluding nvidia-x11-drv-libs-410.78-2.el7_6.elrepo.i686
[nvidia]: excluding nvidia-x11-drv-libs-410.78-2.el7_6.elrepo.x86_64
Installed Packages
dkms-nvidia.x86_64 3:390.42-1.el7 @el7-local
nvidia-detect.x86_64 410.78-1.el7.elrepo @elrepo
nvidia-driver.x86_64 3:390.42-1.el7 @el7-local
nvidia-driver-NVML.x86_64 3:390.42-1.el7 @el7-local
nvidia-driver-NvFBCOpenGL.x86_64 3:390.42-1.el7 @el7-local
nvidia-driver-cuda-libs.x86_64 3:390.42-1.el7 @el7-local
nvidia-driver-devel.x86_64 3:390.42-1.el7 @el7-local
nvidia-driver-libs.x86_64 3:390.42-1.el7 @el7-local
nvidia-libXNVCtrl.x86_64 3:390.42-1.el7 @el7-local
nvidia-settings.x86_64 3:390.42-1.el7 @el7-local
yum-plugin-nvidia.noarch 1.0.2-1.el7.elrepo @el7-local
Available Packages
kmod-nvidia.x86_64 390.87-1.el7_5.elrepo el7-local
kmod-nvidia-304xx.x86_64 304.135-6.el7_6.elrepo elrepo
kmod-nvidia-340xx.x86_64 340.107-2.el7_6.elrepo elrepo
kmod-nvidia-390xx.x86_64 390.87-2.el7_6.elrepo elrepo
nvidia-query-resource-opengl.x86_64 1.0.0-2.el7 el7-epel
nvidia-query-resource-opengl-lib.x86_64 1.0.0-2.el7 el7-epel
nvidia-texture-tools.x86_64 2.0.8-13.el7 el7-epel
nvidia-texture-tools-devel.x86_64 2.0.8-13.el7 el7-epel
nvidia-x11-drv.x86_64 390.87-1.el7_5.elrepo el7-local
nvidia-x11-drv-304xx.x86_64 304.135-1.el7.elrepo el7-local
nvidia-x11-drv-304xx-32bit.x86_64 304.135-1.el7.elrepo el7-local
nvidia-x11-drv-32bit.x86_64 390.87-1.el7_5.elrepo el7-local
nvidia-x11-drv-340xx.x86_64 340.107-1.el7.elrepo el7-local
nvidia-x11-drv-340xx-32bit.x86_64 340.107-1.el7.elrepo el7-local
nvidia-x11-drv-390xx.x86_64 390.87-1.el7_5.elrepo elrepo
nvidia-x11-drv-390xx-32bit.x86_64 390.87-1.el7_5.elrepo elrepo
pcp-pmda-nvidia-gpu.x86_64 4.1.0-5.el7_6 c7-updates

User avatar
TrevorH
Forum Moderator
Posts: 24569
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: Requesting Suggestions: System broken after 7.6 Update

Post by TrevorH » 2018/12/12 17:07:13

You seem to be using packages from nvidia themselves according to what's installed. For example, it says this is installed

nvidia-driver.x86_64 3:390.42-1.el7 @el7-local

Yet further down in the list of available packages we also have

kmod-nvidia-390xx.x86_64 390.87-2.el7_6.elrepo elrepo

No 'elrepo' in the name of the one that's installed so I am only guessing it came from the nvidia repo. But its version number is quite a long way back from the 390.87-2 version from ELRepo that's built for 7.6.
CentOS 5 died in March 2017 - migrate NOW!
Full time Geek, part time moderator. Use the FAQ Luke

tonybKDE
Posts: 11
Joined: 2018/12/06 01:52:48

Re: Requesting Suggestions: System broken after 7.6 Update

Post by tonybKDE » 2018/12/12 17:28:48

OK, that's quite interesting. I can attempt to install those later versions and see what happens.

Questions
Remove the existing 390-42 packages first?
Will dkms-nvidia still automagically do its thing in the future, when the kernel gets updated? I notice there's no dkms-nvidia 390.87 package available.

Post Reply