LM 19.2 broken after system update [solved]

[Petermint]

Small system update in Update Manager. Now everything broken.
LM 19.2 was up to date yesterday. An update today required a rebook then heaps of problems.
Boot startup went from 3 seconds to 3 minutes.

Network Manager is broken. I had the network switched off after the reboot and now I cannot switch it on to check the Internet. Luckily I have a spare computer.

I tried to go back before the update using Timeshift but Timeshift is broken, making Timeshift useless.

I tried to go back one kernal using update manager but update manager is broken the same way as timeshift. When I select the application in the menu, the system waits ten seconds but nothing starts.

Syslog has messages about Flatpak failing. I tried to delete one failing application from the menu but the uninstall link does nothing. Another wait of a few seconds while an icon rotates then nothing.

A brightness control appeared in the applet area after the latest update but the network manager applet was screwed up by the update. Selecting the network manager icon just produces a little vertical bar.

There are many errors on syslog about the power management application.

When an applet fails, surely we should be able to select the real application in the menu!

The big problem with Linux is that you have to fix so many things with the command line. The Terminal application is failing the same as the network manager.

[catweazel]

Small system update in Update Manager. Now everything broken.
LM 19.2 was up to date yesterday. An update today required a rebook then heaps of problems.
Boot startup went from 3 seconds to 3 minutes.

Network Manager is broken. I had the network switched off after the reboot and now I cannot switch it on to check the Internet. Luckily I have a spare computer.

I tried to go back before the update using Timeshift but Timeshift is broken, making Timeshift useless.

I tried to go back one kernal using update manager but update manager is broken the same way as timeshift. When I select the application in the menu, the system waits ten seconds but nothing starts.

Syslog has messages about Flatpak failing. I tried to delete one failing application from the menu but the uninstall link does nothing. Another wait of a few seconds while an icon rotates then nothing.

A brightness control appeared in the applet area after the latest update but the network manager applet was screwed up by the update. Selecting the network manager icon just produces a little vertical bar.

There are many errors on syslog about the power management application.

When an applet fails, surely we should be able to select the real application in the menu!

If you need help, surely you've been around long enough to know to post relevant parts of logs, inxi, etc.

[Petermint]

Most of the diagnostics require terminal which is not working!
I just discovered Ctl-Alt-F1 to start terminal when the terminal application fails.
If I do list diagnostics in terminal, I cannot paste them here because here is a different computer.

Now I am stuck in this other terminal and cannot get back to the regular desktop to browse syslog. Arrrrgghhhh!!!!!!!!!

Good old Windows style reboot. Wait 3 minutes.

[Petermint]

Syslog does not list any errors for network manager application startup failures. It only lists something for power management.

[catweazel]

Syslog does not list any errors for network manager application startup failures. It only lists something for power management.

If you get internet from live media then run it.

[gm10]

I tried to go back one kernal using update manager but update manager is broken the same way as timeshift. When I select the application in the menu, the system waits ten seconds but nothing starts.

The way to go back one kernel is not via Update Manager (unless you're using my PPA, my Update Manager fork lets you select the kernel to boot) but via the GRUB boot menu (press Esc after the BIOS screen before the Mint logo comes up, then go Advanced and select the previous kernel version). You should try this first.

if that helped you can then remove the offending kernel via Update Manager and blacklist that version (right click in the update listing).

If that does not help:

I tried to go back before the update using Timeshift but Timeshift is broken, making Timeshift useless.

Boot the live environment (Linux Mint installation medium), run Timeshift from there, it's in the menu.

[Petermint]

Testing boot from previous kernel.

Last lines in syslog from previous boot:
gvfs-afc-volume-monitor: /usr/lib/gvfs/gvfs-afc-volume-monitor: relocation error: /usr/lib/x86_64-linux-gnu/libgnutls.so.30: symbol _idn2_punycode_decode version IDN2_0.0.0 not defined in file libidn2.so.0 with link time reference
systemd gvfs-afc-volume-monitor.service: Main process exited, code=exited, status=127/n/a
systemd gvfs-afc-volume-monitor.service: Failed with result 'exit-code'
systemd failed to start virtual filesystem service - Apple File Conduit monitor

[Petermint]

Previous kernel did nothing. I will have to download and create live boot.

There were a couple of small updates after the most recent kernel update. One was Mint update manager. The next one added the brightness control and destroyed everything else.

[catweazel]

Previous kernel did nothing. I will have to download and create live boot.

There were a couple of small updates after the most recent kernel update. One was Mint update manager. The next one added the brightness control and destroyed everything else.

You may want to try fsck on the drive.

[Petermint]

Discovered Ctrl-Alt-f7. Such an obvious pair for Ctrl-Alt-F1. Doh!

Download 21 minutes to go then I have to set up a boot device. Will read up on that while i wait.

[basvvliet]

Having the exact same issue. Hope you find any answers!

[Petermint]

Not really solved. Fixed.
I created a LM 19.2 boot usb stick and now store it with the timeshift backup disk.

The problem was a failed systemd update. I restored from timeshift, replacing a few hundred files with almost every file in the systemd directory.

After the restore, I ran the updates one at a time with a timeshift snapshot in between each. In the future, any update to systemd or similar, I will run a timeshift snapshot before and after.

After one of the original updates, there was a message suggesting changing the download server. There were no messages saying anything about errors in the downloaded files or errors applying updates.

After testing tonight's update, I ran a timeshift snapshot. timeshift reports the snapshot is finished before the disk activity light stops blinking, a real worry. When I eject the disk, there is a message about the file system unloading. Nothing changes for a long time. I end up restarting the OS to force a flush of the disk. The OS may sit there for many minutes without shutting down.

When I run rsync direct for a whole of disk backup, rsync ends all the disk activity when it finishes copying. I do not know why timeshift cannot force the flushing out of all writes. It is a real problem with timeshift. Is there a command to force Linux to flush all writes to disk?

I do not know why update manager would recommend changing servers but not detect a download or install error.

[gm10]

Update Manager suggest switching to a different mirror if yours is stale (out of date). A stale mirror should not create issues like yours though. Download or installation errors should be detected.

Given your story about how you are rebooting a system to forcefully end disk activity that's a much more plausible cause of your issues.

[Petermint]

rebooting a system to forcefully end disk activity

The "disk activity" is on the external backup disk, not the system disk where the problem occurred. The backups prove to be 100% ok.
The boot occurs long after the disk activity indicator stops blinking. At that stage it is just Linux Mint displaying a wait message that never goes away.

[gm10]

Got it. I'm not sure what went wrong then. To be exact, a stale mirror could create issues in case of bad packaging (not indicating required version numbers) but I doubt that was the case here.

[fheider]

i had the same problem and solved it really (nice side effect cause i had an issue on my root servers).
https://bugs.launchpad.net/ubuntu/+sour ... ug/1843507

the problem was the ondrej ppa with an issue. libidn2 error.
this destroys the network manager after reboot.

solution:
1 download http://mirrors.kernel.org/ubuntu/pool/m ... _amd64.deb on another pc and copy this file to an usb stick
2 copy file in filemanager to your linux mint system
3 restart mint in recovery mode (boot screen > ESC > advanced options > recovery mode)
4 start root prompt
5 dpkg -i <path>/libidn2-0_2.0.4-1.1build2_amd64.deb
6 reboot

now everything is fine. run an update. ondrej fixed already this bug.

[murray]

I have the exact same problem!

Started up my computer last night and everything was working fine. There was a little update (just a couple of programs but I didn't take a note of what they were), which I installed and then turned the computer off. Turned it back on a few hours later and BLAM! 2 minutes to start up, big delay between logging in and seeing the desktop, no network at all (even the Network icon in Preferences was missing), Terminal wouldn't open (just got a spinning mouse pointer for 20-30 seconds and then nothing), Timeshift wouldn't open (no error message or anything, just nothing happened when I clicked on it).

I rebooted and switched to a previous kernal using GRUB but that didn't help. I plugged in a cable rather than using WiFi but that didn't help. I made a boot disk on a flash drive and used that to boot and everything worked fine, full network and everything, so that told me it wasn't a hardware problem.

I'll try @fheider's suggested fix and see if that works.

[murray]

5 dpkg -i <path> libidn2-0_2.0.4-1.1build2_amd64.deb

Just to clarify, in the dpkg command above there's a space after the path of where I put the .deb file. Is this correct? Eg if I put the .deb file in the directory /home/me/downloads should the command be like this:

Code: Select all

dpkg -i /home/me/downloads libidn2-0_2.0.4-1.1build2_amd64.deb

or should there be no space, like:

Code: Select all

dpkg -i /home/me/downloads/libidn2-0_2.0.4-1.1build2_amd64.deb

I'm guessing it's the latter, no space.

[karlchen]

murray, you are right: no space between the full pathname and the filename.

[karlchen]

Hm. Though it seems that all affected participants in this thread have by now managed to install libidn2-0 version 2.0.4-1.1build2, I do not understand why this libary package arrived on your systems so late?
Checked my dpkg.log files (dpkg.log, dpkg.log.1, dpkg.log.?.gz) and found that on my Mint 19.2, the package libidn2-0 version 2.0.4-1.1build2 arrived end of July, i.e. 6 weeks ago.