mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-06-01 07:01:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 11

fix(docker): dashboard slot stays 'down' when HERMES_DASHBOARD unset

Browse source 
PR #30136 review caught a false positive: when HERMES_DASHBOARD was
unset, the dashboard run script did `exec sleep infinity`, so
`s6-svstat /run/service/dashboard` reported the slot as 'up'.
`hermes doctor` and any other s6-svstat-based health check saw the
dashboard as supervised-running even though no dashboard process
existed.

Add cont-init.d/03-dashboard-toggle: writes a `down` marker file
into `/run/service/dashboard/` when HERMES_DASHBOARD is falsy,
removes any leftover marker when it's truthy. s6-supervise honors
`down` by not starting the service, so s6-svstat reports 'down' —
matching reality.

The run script's HERMES_DASHBOARD case-statement stays in place as
a belt-and-suspenders guard, so the two layers can never disagree.

Two new integration tests lock the behavior: slot reports down
when unset; slot reports up when set to 1.
This commit is contained in:
Ben 2026-05-23 15:24:17 +10:00 • committed by teknium1
parent b28b3f51d3
commit 1dfabe47b3
No known key found for this signature in database
3 changed files with 110 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

55
docker/cont-init.d/03-dashboard-toggle Executable file
View file

@ -0,0 +1,55 @@
#!/command/with-contenv sh
# shellcheck shell=sh
# Toggle the dashboard s6-rc service slot based on HERMES_DASHBOARD.
#
# Runs as root in cont-init.d, after 01-hermes-setup (stage2) and
# 02-reconcile-profiles, BEFORE s6-rc starts user services.
#
# Background (PR #30136 review item I3): the dashboard service was
# always declared as an s6-rc longrun, with its run script checking
# HERMES_DASHBOARD and `exec sleep infinity` when unset. Trouble:
# s6-svstat then reports the dashboard slot as "up" (because sleep
# IS running) even though no dashboard process exists. `hermes
# doctor` and any other s6-svstat-based health check sees a
# false-positive up-state.
#
# Fix: write a `down` marker file into the live service-dir when
# HERMES_DASHBOARD is unset / falsy. s6-supervise honors `down` by
# not starting the service at all, so s6-svstat reports `down` —
# matching reality.
#
# The run script's HERMES_DASHBOARD case-statement stays in place
# as a belt-and-suspenders guard: even if the down marker is
# removed at runtime and the service is brought up, the run script
# still bails when HERMES_DASHBOARD is unset. Both layers agree.
set -eu
# Live service directory for the dashboard longrun. s6-overlay
# compiles /etc/s6-overlay/s6-rc.d/dashboard/ into this location
# at boot, before cont-init.d scripts run.
DASHBOARD_LIVE_DIR="/run/service/dashboard"
# If the live directory hasn't materialized yet (e.g. running in a
# stripped-down test image), nothing to do — the run script's env
# check still keeps things safe.
if [ ! -d "$DASHBOARD_LIVE_DIR" ]; then
echo "[dashboard-toggle] $DASHBOARD_LIVE_DIR not present; skipping"
exit 0
fi
case "${HERMES_DASHBOARD:-}" in
1|true|TRUE|True|yes|YES|Yes)
# Enabled — remove any leftover down marker from a previous boot.
if [ -e "$DASHBOARD_LIVE_DIR/down" ]; then
rm -f "$DASHBOARD_LIVE_DIR/down"
echo "[dashboard-toggle] HERMES_DASHBOARD enabled; removed down marker"
fi
;;
*)
# Disabled — write a down marker so s6-supervise won't start
# the service. s6-svstat will report it as down, matching reality.
touch "$DASHBOARD_LIVE_DIR/down"
echo "[dashboard-toggle] HERMES_DASHBOARD unset; marked dashboard slot down"
;;
esac