apt full-upgrade causes system lock-up

[Bo Berglund]

On Thu, 2 Jun 2022 12:06:22 +0100, Colin Law <clanlaw at gmail.com> wrote:

>> dpkg: unrecoverable fatal error, aborting:
>>  loading files list file for package 'linux-headers-5.4.0-105': cannot read
>> /var/lib/dpkg/info/linux-headers-5.4.0-105.list (Input/output error)
>> Bus error
>> E: Sub-process /usr/bin/dpkg returned an error code (2)
>> W: Problem unlinking the file /var/cache/apt/pkgcache.bin - pkgDPkgPM::Go (30:
>> Read-only file system)
>
>I suspect there is a problem with the hardware.  For some reason it
>has put the disc into read only mode as a protective measure.  If you
>can reboot then first check the SMART data for the disc.  If that
>doesn't show any problems then check syslog to see why it put the
>drive into read only.  However, if there is any important data that is
>not backed up that is really important then you might want to take
>alternative action, such as taking the disc out and recovering the
>data rather than rebooting, in case you lose everything.

In all cases so far (three if I remember correctly) this has hit me during
apt-upgrades and I could do nothing at all via SSH or other means of
communication afterwards.

The first time it sat there until I could visit the site a month later, at which
time I just pulled and reinserted the power line and all was well and dandy.
I could successfully do the apt update && apt full-upgrade too.
No problems then.

I had prepared for the visit and acquired a Shelly Plug which can be controlled
via WiFi to be put on the outlet feeding power to the computer.
So next time it happened I could remotely power cycle it, same today.

Note that the system is apparently fine after the forced power cycle and is
working well with the normal automation tasks.

But now I want to get to the bottom of it, since I wnat to be able to manage
updates via apt...

So I looked for the apt log from today, unfortunately it contains only two
logged commands:

Start-Date: 2022-06-02  07:40:25
Commandline: /usr/bin/apt autoremove

Start-Date: 2022-06-02  07:42:16
Commandline: /usr/bin/apt upgrade

Both were done AFTER the crash and power cycle, and succeeded without problems.

And syslog seems not to contain any line concerning manual start of apt, such as
the command line "apt full-upgrade" or similar so I cannot pinpoint the timing
of the command leading up to the crash.
Syslog itself is too massive to browse through manually using only command line
tools...


-- 
Bo Berglund
Developer in Sweden