Bad UX
systemd
Wiki: wiki/systemd
- Cryptic error codes in
systemctl status
- Example: On Rocky Linux, SELinux prevents system services from running binaries in a user's home directory.
systemctl status
just shows a generic error message and you have to dig through the audit logs to figure it out.- And if the service file has the wrong SELinux type,
systemctl start
will report, e.g.,Failed to restart caddy.service: Unit caddy.service not found.
which is patently not the error.
- And if the service file has the wrong SELinux type,
- Example: With
systemctl --user
, service files cannot haveUser
orGroup
directives.systemctl status
just showsstatus=216/GROUP
with no other explanation.
- Example: On Rocky Linux, SELinux prevents system services from running binaries in a user's home directory.
- The command-line help is not well-organized. For instance,
systemctl SUBCMD --help
prints a message for all subcommands, not just the one you asked for. - Journal logs for
systemctl --user
don't work out of the box (at least not on my Rocky Linux 9 installation); you have to edit a systemd config file and reboot your system. And of course you have to figure this out on your own.
cron
I wrote about some of the problems with cron on my blog. Not all of them are strictly UX problems, although the fact that the job's output goes into oblivion in certain configurations certainly is.