Skip to content

Service install scripts fail silently - #9124

Open
animadito wants to merge 1 commit into
omacom:quattrofrom
animadito:set-e-service-install
Open

animadito wants to merge 1 commit into
omacom:quattrofrom
animadito:set-e-service-install

Conversation

@animadito

@animadito animadito commented Aug 30, 2026 •

Copy link
Copy Markdown
Contributor

Service install scripts fail silently

bin/omarchy-install-service-{dropbox,nordvpn,once,tailscale} don't declare errexit. omarchy-pkg-add failures get swallowed and the script proceeds to enable systemd units and wire up bar entries that lead to nothing.

Related: #9123.

The four bin/omarchy-install-service-{dropbox,nordvpn,once,tailscale} scripts install packages via omarchy-pkg-add, enable systemd units, and set up user-level services. None of them declares errexit, so a failure in any step is silently swallowed and the script moves on. A bar entry can end up enabled for a service that never installed.

The install/hardware/* scripts don't have this problem because they go through run_logged, a bash -eE wrapper. These bin/ entries are exec'd directly by bin/omarchy, with no error wrapper around them.

Each script gets a set -e line directly under its existing metadata header, the same pattern sibling scripts in bin/ already use.
@animadito
animadito marked this pull request as ready for review August 30, 2026 08:48
@omarchybot omarchybot added the bug Something isn't working label Sep 27, 2026
@omarchybot omarchybot added the verified Omarchy Triage has verified that this issue is ready for final review label Oct 1, 2026
@omarchybot

Copy link
Copy Markdown
Collaborator

Reviewed at 81cc815 against quattro. The bug is real and this fixes it.

What was checked. On a disposable Omarchy worker VM, each of the four installers was run with omarchy-pkg-add made to fail and every later command stubbed to record itself. On quattro, all four carried on after the failed install and exited 0: Dropbox enabled its bar plugin and started the client, NordVPN enabled nordvpnd and added the group, ONCE enabled its background service, and Tailscale ran tailscale up, set the operator, enabled the Taildrop receiver, added the bar plugin and installed the web app. With this change, all four stop at the failed install with exit 1. With set -e removed again, the bug comes back. With the install succeeding, every step still runs and each script exits 0. The real omarchy-pkg-add exits 1 for a package it cannot install, so this is the path users hit. ./test/cli passed in full (116 tests) on the worker.

Does set -e break a good install? I read every step after the install. Nothing in a normal run returns non-zero: omarchy-plugin-enable omarchy.dropbox and omarchy.tailscale both returned 0 on the worker, including when the plugin was already enabled. Dropbox's launch runs in the background, and NordVPN's closing gum confirm … && omarchy-system-reboot is an && list, which set -e does not abort on. That line's exit status when the user declines is the same as before. The change also brings these four in line with the other four service installers (1password, signal, spotify, sunshine), which already use set -e.

Second opinion. None. The configured reviewer (Codex Medium) answers, but omabot's daily review budget was used up, so this review is one model's reasoning (Claude Opus 5.5), not two.

Related. #13533 rewrites bin/omarchy-install-service-tailscale and changes the same lines this inserts into, so whichever lands second will need a rebase there.

What happens next. This is labelled verified. Before it can be marked ready, it still needs the second-opinion review and CI to report on this head; nothing is waiting on you.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working verified Omarchy Triage has verified that this issue is ready for final review

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants