Iris found the backend crash-looping by checking rc-service by hand,
because the card could only say `stopped` -- which reads as a state
somebody chose. dev-updater's contract now has a fourth word, `failed`,
and this script implements it.
The OpenRC detail is the one worth not rederiving: it prints `crashed`
*and* exits non-zero, so the word is read rather than the exit code.
Leaning on the code would report "couldn't check", which is a different
and less useful claim. The `running` check stays on its exit code, which
already worked and does not depend on wording.
The script also arranges the logging rather than only reporting it,
because neither unit wrote a file: systemd went to the journal and
OpenRC's `command_background=true` discarded output entirely, which is why
a crash left nothing to read. Output now goes to
$XDG_DATA_HOME/ai-server/ai-server.log -- generated data, outliving any one
build, and not in a repository shared with a machine that should not read
it. `start` rotates one generation aside, so what is kept is exactly this
run and the one before: the pair worth having after a crash and a restart.
`logs` prints the paths, newest first, and nothing else.
Verified on systemd by causing the failure rather than reasoning about it:
installed, started, confirmed `running` on the wg0 bind, wrote an
unparseable config, restarted, and watched status settle on **failed**
rather than stopped -- with the reason, line and column, in the file `logs`
points at, and the crash preserved in .1 after recovery. Then restored,
confirmed `running` again, and uninstalled.
**The OpenRC branch is written from the documentation and is untested**,
here and in dev-updater, since neither machine that can run it is one
either of us can test on. It is also the branch that actually matters,
since the backend runs under OpenRC on the host. `output_log`/`error_log`
in the openrc-run script are the parts to distrust first.
One thing that bit while writing it: the systemd heredoc is unquoted so
$LOG expands, which makes a backtick in a comment inside it run as command
substitution. A comment saying "`start` rotates" executed `start`, and the
unit was written without ever being valid. There is now a note in the
heredoc saying why it contains no backticks.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017xn8nHw1tw1R6PtiY1eEtw
Dev Updater's unit is a project -- a directory in a checkout -- that
produces components, so this declaration belongs at the repository root
where it can say where each half lives. `cwd` is how a component says
that, and it is what the old file was working around: sitting in `app/`,
it reached the backend with `../server/Cargo.toml` and `../server/service`,
which described the layout backwards.
So the file says what this repository produces: the backend, built and
serviced in `server/`, and then the APK, built in `app/`. The order is
unchanged and still the point -- a failing APK build leaves the phone the
APK it already had rather than half of a matched pair.
The service script gains dev-updater's note about lingering. A user
service stops at logout unless `loginctl enable-linger` is set, which for
this one matters more than for a build server: the phone reaches ai-server
whether or not anyone is logged in at the desk.
**This changes the project path, so the phone's existing entry (at
~/repos/ai-app/app) has to be removed and the checkout root added
instead**, and its components accepted once.
Verified against dev-updater's current parser rather than by reading its
schema: the declaration loads, `cargo build --release` resolves into
server/, app/build-apk.sh into app/, the service script to an absolute
path, and APK discovery from the root still finds the built APK.
`./server/service status` prints not-installed and exits 0.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017xn8nHw1tw1R6PtiY1eEtw
ai-server is now something Dev Updater can install, start and stop from
the phone: the project declares a Server component naming ./server/service,
and that script is where knowing about systemd and OpenRC lives.
Detection asks rather than looks -- `systemctl --user show-environment`
answering, or `rc-service --user` existing -- because a machine can carry
both binaries and an OpenRC below 0.60 has rc-service without --user.
Neither present is an error with a message, not a guess at a fallback.
Two things the script will not do. It never prompts: Dev Updater runs it
with stdin closed, so a sudo prompt would hang rather than fail, and
anything needing root exits telling you to run it yourself once. And on
OpenRC it refuses when XDG_RUNTIME_DIR is unset rather than proceeding --
user services store their state there, and the failure otherwise reads as
a broken service instead of a missing variable.
The server is declared before the app so it is built and delivered first;
a failed APK build then leaves the phone with the APK it already had
rather than half of a matched pair.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QQz6R4kBQcWSHBNZMgBnjL
irisandClaude Opus 5