echo "$list" | grep -q under set -o pipefail exits 141 whenever grep
matches before echo finishes writing, so healthy browsers were reported
'CDP up but no page target' and killed every 2 min fleet-wide (load 19+).
Use [[ == *glob* ]] (no pipe, no race) for the page/muse.ai stages, and
grep -c (reads to EOF, never early-exits) for the netns check.
The relaunch-loop guard only protected the kill step, not the relaunch.
When CDP was unreachable on a slow-starting browser, the watchdog would
invoke netvm-chrome.sh (which kills the existing browser) before checking
if it was recently launched — piling up 5 chromiums on opm.
Now the <120s check runs before the relaunch and skips the entire cycle.
Also fixed the PID check to match only the main browser process
(--remote-debugging-port, excluding --type= renderer/gpu children).
Session: sidechat/chromebox-ops
- Added CHAT_POLICY.md and README.md banner enforcing sidechat-first and file-transfer-first rules.
- Added strict Main Chat block to super dm send and super dm wo with --allow-main-chat override.
- Implemented file transfer staging and metadata registry in super dm send-file and super dm files (with clean subcommand).
- Added full job lifecycle management (show, create, enable, disable, delete, run --follow) to super-cli.py and box-ctl.py.
- Audited all jobs in jobs/*.json and redirected automated dispatches away from Main Chat.
- Hardened chromebox-watchdog.sh with systemd user session environment exports and stale singleton cleanup.
- Added compose_check command choice to muse-chat-api.py.
Two fixes: (1) retry health check 4x with 15s gaps after relaunch instead of single 25s check; (2) skip kill if browser launched <2min ago (probably still starting). Prevents watchdog from killing a working-but-slow browser.
Session: sidechat/chromebox-ops
Root cause: agent-health.service runs Type=oneshot with the default
KillMode=control-group. restart_browser() spawned the replacement
chromium with nohup under the service, so systemd SIGKILLed it the
moment the service exited. Every 5-min tick: FAIL -> restart ->
RECOVERED -> SIGKILL at teardown. opm and pip were permanently dark
and dm.py read masked it (empty output, exit 0).
Fix: launch replacements via systemd-run --user --scope (backgrounded)
so the browser lives in a transient scope outside the service cgroup
and survives teardown. setsid does NOT escape either. Note:
systemd-run --scope waits for the scope even with --no-block
(verified 2026-10-03), hence the backgrounding. Same fix in
chromebox-watchdog.sh (timers currently off).
dm.py: dm_read() now uses run_full() and prints a WARNING to stderr
with rc + last error line instead of failing silently on empty reads.
Trailers: Session: sidechat/opm-blind-fix
operator