provision script failure - assumed /root path

[ITJamie]

/srv/osctrl# ./deploy/provision.sh -m prod --all-hostname "osctrl.domain.tld" --type own --nginx --postgres --redis -p all -E
./deploy/provision.sh: line 396: /root/osctrl/deploy/lib.sh: No such file or directory

there is a hard assumption that the repo will be cloned in /root instead of another path

[alvarofraguas]

Repro confirmed. The default SOURCE_PATH is the issue.

Root cause

deploy/provision.sh:164 defaults SOURCE_PATH to a fixed string:

SOURCE_PATH=~/osctrl

When the script runs as root, ~ expands to /root → SOURCE_PATH=/root/osctrl. If the clone lives anywhere else (/srv/osctrl, /opt/osctrl, etc.) and the caller doesn't pass -s /actual/path, the next reference at line 396 (source "$SOURCE_PATH/deploy/lib.sh") fails with No such file or directory.

The -s PATH option exists but is easy to miss in the help block — and it shouldn't be required when the script is being executed from the clone itself.

Suggested fix

Replace the hardcoded default with a script-relative path detected at startup:

SOURCE_PATH="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"

(That sets SOURCE_PATH to the parent of the directory provision.sh lives in — i.e. the repo root, regardless of where the clone sits.)

Behaviour deltas:

• Operator runs ./deploy/provision.sh ... from /srv/osctrl/ → SOURCE_PATH=/srv/osctrl automatically. ✅
• Operator runs from /root/osctrl/ (the historical default) → SOURCE_PATH=/root/osctrl still. No regression.
• Operator passes -s /custom/path → still honored, override semantics unchanged.

Two-line change to provision.sh. Happy to PR it if useful.