Handle database on inaccessible NFS share gracefully

sburris0 commented 1 year ago

Overview

When a keepass db is on an inaccessible NFS share, it hangs rather than handling the error gracefully.

Steps to Reproduce

Setup:

Set up an NFS share with your DB on it
Mount the share from another machine
Start keepassxc, unlock DB (so it saves DB location)
Close keepassxc
Actual reproduction:
Make the NFS share inaccessible (e.g. disable wireless, unplug ethernet, etc)
Start keepassxc
Observe that it hangs/is a black screen

Expected Behavior

Upon starting keepassxc with an inaccessible DB file, I expect:

keepassxc does not hang or crash
in the dropdown, the inaccessible db is grayed out/disabled. Probably with some sort of explanatory label like "(Inacessible)"

Ideally, an inaccessible DB should not impair the function of keepassxc. "Display an error and exit" is a bad strategy because the user may have other DBs that work.

Actual Behavior

Keepassxc starts and the window is entirely black. After a few minutes, I am asked if I want to Force Quit the application.

Context

My use case is storing a DB on my NAS, mounting the NFS share from my workstation, and using keepass from there.

$ keepassxc --debug-info # can't start GUI right now
Warning: Ignoring XDG_SESSION_TYPE=wayland on Gnome. Use QT_QPA_PLATFORM=wayland to run on Wayland anyway.
KeePassXC - Version 2.7.4

Qt 5.15.8
Debugging mode is disabled.

Operating system: NixOS 23.05 (Stoat)
CPU architecture: x86_64
Kernel: linux 6.0.11

Enabled extensions:
- Auto-Type
- Browser Integration
- SSH Agent
- KeeShare
- YubiKey
- Secret Service Integration

Cryptographic libraries:
- Botan 2.19.3

Operating System: Linux (Nixos) Desktop Env: Gnome Windowing System: Wayland

droidmonkey commented 1 year ago

I can never reproduce this problem on Windows or Ubuntu. It is either a poor interaction with Qt and your desktop env, or it's the desktop env itself not properly reporting inaccessible file.

sburris0 commented 1 year ago

I wonder if it's GVFS-related. I noticed that stat <db file> doesn't error as expected but also hangs

droidmonkey commented 1 year ago

Right so it isn't really on us. Although we should enforce a time out somehow. Unfortunately qt doesn't have that built into their file api. Really GVFS needs to properly handle this situation.

sburris0 commented 2 months ago

I've just run into this again. This time running sway wm rather than GNOME. ps output:

spencer   327274  0.1  0.2 1476992 42564 ?       Sl   16:42   0:00 foot
spencer   327275  0.0  0.0   9384  5888 pts/4    Ss   16:42   0:00  \_ /run/current-system/sw/bin/bash
spencer   329684  2.1  0.0  10312  6912 pts/4    S    16:51   0:00      \_ bash --rcfile /tmp/nix-shell-329684-0/rc
spencer   332724  0.2  0.5 674592 89576 pts/4    DLl+ 16:51   0:00          \_ keepassxc

I don't think there is GVFS in the picture here:

[spencer@nixos:~]$ systemctl cat mnt-nas.mount
# /run/systemd/generator/mnt-nas.mount
# Automatically generated by systemd-fstab-generator

[Unit]
Documentation=man:fstab(5) man:systemd-fstab-generator(8)
SourcePath=/etc/fstab
Before=remote-fs.target

[Mount]
What=192.168.2.32:/tank/data
Where=/mnt/nas
Type=nfs

I do have local copies of my DB, but I can't get keepassxc to start to access them. It would be nice to have some sort of failsafe, like a CLI flag that tries to access ONLY databases passed as an argument, or one that ignores the cache

keepassxreboot / keepassxc