Understanding Systemd Units and Unit Files

This comment has been deleted

Your article is the best I have read on the subject. The only addition I see would be about the ‘systemctl show unitFile’ that displays all the configuration details about a unit file.

hi! nice article! Is it possible to make a .service to wait a HDD gets fully mounted before exec it? Today I add a sleep in pre start unit. I don’t use fstab. My O.S auto-mounts the HDD. I didn’t set anything. Tried add the media-HDD.mount in “After” but didn’t work. thanks!

Thanks for your article!

really helpful!

These tutorials are well written and super helpful. Thanks a lot!

Great article.

Formatting nitpick: in Types of Units, .service is missing a bullet.

The following locations should also be known for .service files:

Runtime

• Runtime units: $XDG_RUNTIME_DIR/systemd/user

User (when using “systemctl --user”)

• User Unit Files: $XDG_CONFIG_HOME/systemd/user
• User Unit Files (when $XDG_CONFIG_HOME is not set): $HOME/.config/systemd/user

Override

• Override Unit File Load Path: $SYSTEMD_UNIT_PATH

Reference: https://unix.stackexchange.com/a/367237

Very informative and deep article. Thanks.

Maybe it could be interesting to add something about how to run operations on services status, exploring also commands like systemd-delta and systemctl daemon-reload.

And maybe some real example at the end and some picture to have a fast re-read of the article, like for the unit files hierarchy :)

Best Regards

You left out %j, the part of %p after the last hyphen (if there is one, otherwise it is equal to %p). And %J of course.

I think it sad that systemd doesn’t have any string utilities to manipulate %I. E.g. if you pass two arguments as instance: “arg1 arg2” then there is no way to add a rule: After: foo@arg1.service because the only way to get to arg1 is with scripting (aka, as part of a ExecStart). And because of that, there is the question of how to mimic all of the other rules (like Wants:, Requires:, After: etc) with an ExecStart:…