Update Failure

Hi all,

just registered for my very first post - didn’t have any issues in the past. :wink:
Should I accidentally disregard any forum rules or best practices, please teach me.

I have no understanding of the actual issue, so I’ll give anything I have until now. I’m happy to provide additional information on request.

Today’s Plasma Update hung when trying to install Avahi (or so it said). After a restart, it will notify me:

Failed to activate service ‘org.freedesktop.PackageKit’: timed out (service_start_timeout=25000ms)

I cannot resort to zypper as it tells me:

zypper: error while loading shared libraries: libzypp.so.1700: cannot open shared object file: No such file or directory

There are, however, plenty of those under /.snapshots, and a symlink /usr/lib64/libzypp.so.1702 pointing to libzypp.so.1702.0.2 beside it.

YaST will complain:

Internal Error. Please report a bug report with logs.
Run save_y2logs to getcomplete logs.
Details: component cannot import namespace ‘Pkg’
Caller: /usr/lib64/ruby/vendor/ruby/2.5.0/yast/yast.rb:178 in ‘import_pure’

This is the same for the majority of components that I try to start from YaST: In the end it’ll come down to the very same Details and Caller.

I much appreciate any help to get my system straight. Thx!

start yast ncurses and proceed to complete the update

sudo yast2 --ncurses

Well, /usr/lib64/libzypp.so.1702 is not the same as libzypp.so.1700… (actually the change in the number means that they are incompatible)

Your system is in an inconsistent state because of the “aborted” update.
Unfortunately, zypper is broken by that, so it won’t be easily possible to fix it. (you’d need to download affected packages manually and install it with rpm one by one)

I would suggest one of two things to repair your system:

  • boot from an installation medium and do an “Upgrade”, this will basically reinstall all packages.
  • boot an earlier snapshot and roll back. (“snapper rollback”)

And in the future, you should only use “zypper dup” to update your Tumbleweed system.

Likely not easy to do, as YaST fails to start as well… :wink:

Well, YaST itself starts, but the update component won’t with the above error.

Usually yast ncurser also starts when Yast Gui does not work, other ideas I have not seen that the author of the post says that yast gui ne zypper work, but using the translator (also to remember what I wrote) maybe I lose something unfortunately

Try

rpm --rebuilddb

That depends on why the YaST Gui doesn’t work.

In this case:

Internal Error. Please report a bug report with logs.
Run save_y2logs to getcomplete logs.
Details: component cannot import namespace 'Pkg'
Caller: /usr/lib64/ruby/vendor/ruby/2.5.0/yast/yast.rb:178 in 'import_pure'

which will also affect the ncurses version.

And “rpm --rebuilddb” won’t help here either. This is definitely not a problem with the RPM database, but with the installed packages.

I’m aware of that fact. I tried to hint on the system’s state.

Thx, I feared there’d be no simple fix…
Could you please be more verbose esp. on the 2nd alternative, and/or point me to howtos?

Okay… The notification itself does no harm and may remain active?

https://doc.opensuse.org/documentation/leap/reference/html/book.opensuse.reference/cha.snapper.html#sec.snapper.snapshot-boot

Basically you choose “Start bootloader from a read-only snapshot” and the boot menu, select a snapshot to boot, and then run “snapper rollback” as mentioned.

Okay… The notification itself does no harm and may remain active?

Correct.

Although I’m not completely sure that this was actually the problem here, as you say that it hang during the installation of one particular package.
Maybe that would have happened with zypper as well… :\

On 03/17/2018 12:46 PM, wolfi323 wrote:
>
> Zaphod_B;2859006 Wrote:
>> I cannot resort to zypper as it tells me:
>>>
> Code:
> --------------------
> > > zypper: error while loading shared libraries: libzypp.so.1700: cannot open shared object file: No such file or directory
> --------------------
>>>
>> There are, however, plenty of those under /.snapshots, and a symlink
>> /usr/lib64/libzypp.so.1702 pointing to libzypp.so.1702.0.2 beside it.
>>
> Well, /usr/lib64/libzypp.so.1702 is not the same as libzypp.so.1700…
> (actually the change in the number means that they are incompatible)
>
> Your system is in an inconsistent state because of the “aborted” update.
> Unfortunately, zypper is broken by that, so it won’t be easily possible
> to fix it. (you’d need to download affected packages manually and
> install it with rpm one by one)
>
> I would suggest one of two things to repair your system:
> - boot from an installation medium and do an “Upgrade”, this will
> basically reinstall all packages.
> - boot an earlier snapshot and roll back. (“snapper rollback”)
>
> And in the future, you should only use “zypper dup” to update your
> Tumbleweed system.
>
>

I further recommend using ‘zypper dup --download-only’ until all
packages are downloaded as it eliminates problems during the downloading
portion is complete. After the downloads are complets then use ‘zypper
dup’ to finish the process.


Ken
linux since 1984
S.u.S.E./openSUSE since 1996

I believe you can use the installation DVD to update
https://download.opensuse.org/tumbleweed/iso/openSUSE-Tumbleweed-DVD-x86_64-Snapshot20180314-Media.iso

Thx again to all of you. I’ll look into your suggestions and get back to you if (hopefully) my system is in a sane state again or I get stuck in the process.

Should I file a bug report as YaST suggests, or is that misleading?

If I select a snapshot to boot into (I tried the three most recent), after hitting “if ok run snapper rollback and reboot” or the like I get

error: null src bitmap in grub_video_bitmap_create_scaled

Press any key to continue...

and doing so brings me back to the previous screen.

I’d really like to follow the snapshot path for ease. Any advice?

Hm, sounds like a problem with the grub files (in the snapshot?)

Well, maybe an older one would work?

Or, as it seems to be related to some graphic bitmap, I could imagine it may help to switch the boot menu to text mode, by setting this in /etc/default/grub:

GRUB_GFXMODE=text

And then run “sudo grub2-mkconfig -o /boot/grub2/grub.cfg” to regenerate the menu.

Otherwise, I’m not sure what that message could mean or how to “fix” that problem (if even possible), sorry.

But just to avoid a possible misunderstanding:
The message “if ok run snapper rollback and reboot” is just informational, you shouldn’t try to select it (though I’m not sure that’s even possible…)

I’d really like to follow the snapshot path for ease. Any advice?

Well, an “Upgrade” installation (booting from the installation DVD or CD and choosing “Upgrade”) should be quite easy as well. As I wrote, it would basically just reinstall all packages and not touch your system otherwise.

Another thing you could try is install the latest zypper manually, and then run “zypper dup” again to finish the update, in the hope that zypper will work then.

sudo rpm -Uvh http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/zypper-1.14.4-1.1.x86_64.rpm

I once got into a similar situation. Here’s what I did, and what might perhaps get things up and running.

  1. Boot into a rescue system – either the install DVD in rescue mode, or a live Tumbleweed;
  2. Mount your root partition at “/mnt”. Hmm, if it is “btrfs” you also have to mount subvolumes.
  3. Mount everything else relevant (such as “/boot” and/or “/boot/efi”).

mount --bind /dev /mnt/dev
mount --bind /proc /mnt/proc
mount --bind /sys /mnt/proc

  1. Do an update for the mounted system. But run zypper itself in the rescue system
zypper --root /mnt dup

The idea is that you are using the working zypper in the rescue system (the booted system) instead of the broken zypper in the mounted system.

Shouldn’t be necessary.
zypper normally does download all packages before installing anything, unless you changed the defaults.

Opted for this and, at first sight, seems to have worked.

Had to resolve some conflicts between Packman and openSUSE repos. Also was shocked by a digest mismatch of the kernel download, but this got resolved by a zypper refresh.

Hope the system is still in sane state after the next updates rolled in…

Thx again for the quick and to-the-point help. My first experience with this forum has been outstanding!:good:

Yes, which Wolfi already said.

Never leave an update in half, if yast or zypper crashes exit (retry- ignore- skip) and never restart without retrying the update

You’ve been lucky that wolfi323 knows one more than the devil

After updating yesterday I wound up with a desktop and no applications that would start. I am aware of the preference for “zypper dup” but I have, by and large, not had any problems starting the update from the panel icon. So I ran “zypper dup” and got numerous reports about packages that were conflicting. The problem is the options for resolving the package conflicts are meaningful to experts, but leave the “middle class” Tumbleweed user at a loss. Anyhow, somehow I got my Tumbleweed working again, although I would rather know more about the impact of the options on the system’s functionality.

By “middle class” Tumbleweed user, I mean I have used SuSE since 6.1 and also run Debian derivatives on a multi-boot system. So I am not a “newbie” but neither am I a programmer.