On a systemd host the logs are in a binary journal rather than plain files, which sounds like a step backwards until the first time you need one service’s output for a fifteen-minute window and do not have to work out which file it went to.
journalctl -u nginx --since '15 min ago'
journalctl -u queue-worker -f # follow
journalctl -u queue-worker -p err # errors and worse
journalctl -u queue-worker --since yesterday --until '09:00'Filtering by unit, priority and time range without a single grep is the whole argument, and it works the same on every systemd host regardless of where a package decided to write. The journal is not persistent by default on some distributions — it lives in /run and disappears on reboot until /var/log/journal exists. Creating that directory is usually the first thing worth doing.