Hi,
based on my understanding sysexts are geared towards scenarios where either other services need to consume your files or you are able to consume files by the OS, e.g. system libraries. You can for example build these using mkosi which uses your normal distribution packages.
Portable services on the other hand run complete in their own filesystem and do not use files from your default rootfs. This is ideal for services which are statically compiled or if your service is usually only packaged for e.g. Debian but you want to run it on a Fedora host.
Cheers, Nils
On Sat, Jan 25, 2025, 19:00 Sopena Ballesteros Manuel <manuel.sopena@xxxxxxx> wrote:
Dear systemd community,I’m a sysadmin currently learning Linux and systemd, and I’m exploring technologies that allow me to add and remove applications to my system’s root filesystem in a plug-and-play fashion. I’m avoiding containers because I need a higher level of integration with the system, and the abstraction containers provide doesn’t align with my needs.As I learn more about systemd, I came across systemd portable services and systemd-sysext while researching solutions to my problem. From my understanding, sysext leverages OverlayFS, while portable services seem to function more like a chroot environment with additional functionality. Although they appear to have similar use cases, I’m having trouble deciding between the two.Could anyone provide clarification on the specific scenarios where one solution would be more appropriate than the other?Thank you very much for your time and assistance!Best regards,
