Strange flatpak-wide freezing after attempting SSH-tunnelled MariaDB database dump

Summary: My system would partly freeze after attempting my first MariaDB SSH-tunnelled remote DB dump using the Flatpak version of DBeaver. The backup didn't seem to initialize properly, the 'freezing' affected all running flatpak apps and prevented any new ones from starting, and I further believe the issue is in some way related to flatpak as the non-packed version of the app doesn't appear to exhibit the same behaviour (the backup works, no freezing).

--

I encountered this issue after setting up an SSH-tunnelled connection to a MariaDB server and attempting my first database dump using this package. After providing it with the path of 'mysqldump' (symlinked to 'mariadb-dump'), it would seemingly start the backup but report/make no progress in terms of data transfer, freeze the app for about 20-30 seconds, and then partly unfreeze to the point that I could close some of the backup windows and get back to the main client which was now unresponsive.

It then triggered some weird system instability, symptoms follow:

- most process/resource monitoring apps and commands would silently refuse to run. Top, atop, system monitor, resources, mission center would refuse to start, silently, and some of them would crash if already running when the backup was started. Thankfully while 'top' wouldn't start after a backup-freeze, it would keep running if it had been started before-hand.
- Any other flatpak apps open when the backup started would either immediately crash, or freeze and then eventually crash.
- I was not able to start/restart any flatpak apps, they would silently fail to load.
- Attempting to cleanly shut down / restart the system also failed silently.

Looking at 'top' while one of these instability periods occurred didn't reveal anything obviously wrong ... no processes spiralling out of control or resources being unexpectedly consumed. I went around restarting services and killing processes to see if that helped, and I discovered that if I terminated the 'flatpak-session-helper' process, I would be able to start one flatpak app successfully, and then the next one I attempt to start would freeze, unless I once again terminated 'flatpak-session-helper'.

After a while of not seeing any indication of what was causing the problem, apps silently failing, being unable to shut down or restart the system, no obvious errors in system logs, and no sign of a solution, I would close the non-flatpak desktop apps that were - up until this point - still running normally, wait a small amount of time for dust to settle, and then reluctantly do a hard restart.

After a few rounds of this, I uninstalled the Flatpak DBeaver client and installed the arch package. Since then, no sign of this issue ... though I'd be willing to reinstall the flatpak to replicate the issue if there are specific tests you wish me to run.

Background:
CachyOS (Gnome) - Kept up-to-date dailyish
DBeaver version: 24.3.3

Notes:
Between uninstalling the flatpack and installing the arch package, the settings persisted and connections were still in place. I had to change the DB dump local client to use a native system path rather than a flatpak run path, but otherwise the same settings were used.

Start by reproducing an SSH-tunnelled MariaDB dump in the Flatpak DBeaver client using the reported mysqldump/mariadb-dump path, while monitoring flatpak-session-helper and the affected applications. Compare it with the Arch package and inspect system behavior during the dump; done means the backup completes without freezing DBeaver, other Flatpak apps, or system shutdown.

Written by the indexing model from the issue text.