Conversation

Notices

1. Embed this notice
   viq (viq@social.hackerspace.pl)'s status on Sunday, 03-Nov-2024 21:22:03 JST viq

   Annoying thing with #flatpak apps is that their temporary files go to `$XDG_RUNTIME_DIR`, which is a ramdisk created by systemd. Ramdisk created with very limited size, causing for example #Signal to fill it, or things to quite comically fail when #calibre tries to transfer a larger book to your device. And I haven't yet found a way to adjust the size of this ramdisk, and only an ugly hack to work around this.
   #Linux #sysadmin

   • Embed this notice
     Solène :flan_hacker: (solene@bsd.network)'s status on Sunday, 03-Nov-2024 21:22:02 JST Solène :flan_hacker:

     in reply to

     @viq does it work if you set TMPDIR in the user systemd environment?

   • Embed this notice
     Solène :flan_hacker: (solene@bsd.network)'s status on Sunday, 03-Nov-2024 21:29:39 JST Solène :flan_hacker:

     in reply to

     @viq edit a file like

     ~/.config/environment.d/user_vars.conf

     and put TMPDIR=/something/that/exist

     maybe it will work for your issue 🤔 , the default value of TMPDIR is /tmp

   • Embed this notice
     viq (viq@social.hackerspace.pl)'s status on Sunday, 03-Nov-2024 21:29:41 JST viq

     in reply to

     • Solène :flan_hacker:

     @solene sorry, where/how do I set that?

   • Embed this notice
     Solène :flan_hacker: (solene@bsd.network)'s status on Monday, 04-Nov-2024 03:05:49 JST Solène :flan_hacker:

     in reply to

     @viq maybe it's XDG_RUNTIME_DIR

     see https://github.com/flatpak/flatpak/issues/3438

   • Embed this notice
     viq (viq@social.hackerspace.pl)'s status on Monday, 04-Nov-2024 03:05:50 JST viq

     in reply to

     • Solène :flan_hacker:

     @solene with this in place at least Signal still puts the `.org.chromium.Chromium.XXXXXX` files in `/run/user/<uid>`

   • Embed this notice
     viq (viq@social.hackerspace.pl)'s status on Monday, 04-Nov-2024 03:05:52 JST viq

     in reply to

     • Solène :flan_hacker:

     @solene thank you. I'll try that later, though flatpak has some assumptions and isolations.
     It probably has a better chance of working for calibre than IMs, and for that someone suggested also other variables to try, specific to calibre