Finish an update that a migration aborted with 'could not put ... on the bar: omarchy-shell is not responding'
Stable 4.0.4-1 omarchy-shell wraps calls in timeout --kill-after=1s ${OMARCHY_SHELL_IPC_TIMEOUT:-2s} and maps 124/137 to omarchy-shell is not responding. omarchy-migrate runs each migration with bash -euo pipefail and touches the marker only afterwards, and omarchy-update runs migrate under set -e before hooks, AUR, mise, orphans and restart, so the skipped-steps claim holds. Neither migration is in /usr/share/omarchy/migrations here (newest 1789444024). On quattro, bin/omarchy-bar ask_to_put still retries only not ready and is not running and fails on anything else, and migrations/1790042972.sh still calls omarchy-shell -q shell rescanPlugins then omarchy-bar put omarchy.elsewhen with no || true. 1789581661 no longer exists on quattro, which does not affect the record since it quotes reports. #12777 (open, r2186 confirmation, widget already in shell.json, restart plus retry completed) and #13301 (open, stable to edge r2304, repeats every run) support the cause and both workarounds. The environment variable is inherited through script, the lock wrapper, migrate and omarchy-bar to omarchy-shell. Not exercised: no update or shell restart was run.
Do not run omarchy restart shell over ssh or on a locked session: it re-locks the session and the Omarchy 4 lock screen cannot be released headlessly.
Symptom
omarchy update aborts during migrations, after the package step succeeded:
Running migration (1789581661)
Install Elsewhen, the world clock plugin
...
omarchy-bar: could not put omacom.elsewhen on the bar: omarchy-shell is not responding
Something went wrong during the update!
Re-running fails at the same migration, sometimes several times in a row. Migration 1790042972 fails the same way. Often seen when switching from stable to edge.
Cause
omarchy-shell wraps every IPC call in timeout ${OMARCHY_SHELL_IPC_TIMEOUT:-2s} and maps a timeout to omarchy-shell is not responding. ask_to_put in omarchy-bar retries 50 times when the shell is starting (not ready) or absent, but treats a timeout as fatal on the first attempt. During an update the shell is busy reloading the QML the same transaction just replaced, and the migration calls rescanPlugins on the line before the put, so the put often outlasts 2 seconds even though the shell does write the widget into shell.json. The migration dies under set -euo pipefail, omarchy-migrate writes no completion marker, and every later step (hooks, AUR, mise, orphans, shell restart) is skipped. Reported on omarchy-dev r2186 and r2304. Neither migration is in the stable 4.0.4-1 tree here.
Fix
Give the shell IPC more time for this run. The variable is inherited through script and omarchy-migrate down to omarchy-shell:
OMARCHY_SHELL_IPC_TIMEOUT=15s omarchy update
Alternatively restart the shell first from your unlocked session, give it a few seconds to settle, then retry:
omarchy restart shell
sleep 10
omarchy update
The migrations are idempotent, so a retry finishes them. Check that nothing is left pending:
ls ~/.local/state/omarchy/migrations/ | sort | tail -3
ls /usr/share/omarchy/migrations/ | sort | tail -3
Verify
The update ends without the red banner and prints omacom.elsewhen is on the bar (or the relevant widget), and the newest migration file name in /usr/share/omarchy/migrations/ also exists in ~/.local/state/omarchy/migrations/.