fix: auto-restart dashboard after update when argv is recoverable by Variable85 · Pull Request #41508 · NousResearch/hermes-agent

Variable85 · 2026-06-07T20:48:29Z

Summary

Preserve recoverable dashboard argv before stopping stale dashboard processes during hermes update
Auto-restart recovered dashboards after update, adding --no-open for headless/SSH safety
Restart systemd-managed dashboard services through systemctl restart <unit> instead of spawning an unmanaged clone that races the service for the port
Keep the manual restart fallback when argv recovery is unavailable or restart fails
Harden stale dashboard detection against shell-wrapper false positives and quoted Windows executable paths

Test Plan

git diff --check
python -m py_compile hermes_cli/main.py tests/hermes_cli/test_update_stale_dashboard.py
uv run --with pytest --with pytest-xdist --with pyyaml python -m pytest tests/hermes_cli/test_update_stale_dashboard.py tests/hermes_cli/test_dashboard_lifecycle_flags.py -q -o 'addopts=' → 44 passed in 0.88s
Runtime verification on a Linux host with hermes-dashboard.service: /proc/<MainPID>/cgroup resolves to hermes-dashboard.service; helper returns hermes-dashboard.service
Runtime verification: local dashboard stayed active and curl http://127.0.0.1:9119/login returned dashboard HTML

Notes

LAN/public binds still require dashboard auth after the June 2026 hardening; this PR handles restart behavior, not credential provisioning.

Preserve recoverable dashboard argv before stopping stale dashboard processes, then restart those dashboards after update while retaining the manual fallback when argv recovery or restart fails. Harden stale dashboard detection against shell-wrapper false positives and cover quoted Windows paths.

teknium1

Thanks for tackling the update/dashboard mismatch; current main still stops stale processes and leaves a manual restart hint (hermes_cli/main.py:6094-6210), so the underlying problem is valid.

Problems

The new matcher only recognizes dashboard (hermes_cli/main.py:5649 in this PR). Current main also intentionally reaps hermes serve backends (hermes_cli/main.py:5881-5886), so this would regress stale serve cleanup.
The systemd path ignores user.slice (hermes_cli/main.py:5799; see the explicit expectation in tests/hermes_cli/test_update_stale_dashboard.py:507-510). It then falls back to detached spawning, which can race a user service configured with Restart=always. Existing gateway service support has explicit systemctl --user handling in hermes_cli/gateway.py:1999-2002.
Restart output is sent to Path.home() / '.hermes' at hermes_cli/main.py:5857, bypassing the active profile's HERMES_HOME.

Suggested changes

Preserve serve matching and test it.
Handle user- and system-scoped units via the scope-aware service helpers instead of treating user units as unmanaged.
Use get_hermes_home() for the restart log path.

Automated hermes-sweeper review.

teknium1 · 2026-07-14T05:45:01Z

+        # `python -m hermes_cli.main dashboard` and `hermes dashboard` both
+        # place the subcommand after this token.  Permit global flags before
+        # the subcommand (e.g. `hermes --profile taste dashboard`).
+        if "dashboard" in tail:


This drops current-main support for hermes serve: main currently treats hermes serve, hermes_cli.main serve, and hermes_cli/main.py serve as stale backends too (hermes_cli/main.py:5881-5886). Preserve that command family here and add a regression test before replacing the existing matcher.

teknium1 · 2026-07-14T05:45:01Z

+                # cgroup v2: 0::/system.slice/hermes-dashboard.service
+                # cgroup v1/systemd: 1:name=systemd:/system.slice/foo.service
+                path = line.rstrip("\n").split(":", 2)[-1]
+                if not path.startswith("/system.slice/"):


A per-user systemd service appears under user.slice, so this branch makes it fall through to detached Popen rather than service restart. That can race a Restart=always unit. Please detect user scope and use systemctl --user; hermes_cli/gateway.py:1999-2002 already centralizes its environment setup.

teknium1 · 2026-07-14T05:45:01Z

+
+    restarted: list[list[str]] = []
+    failed: list[tuple[list[str], str]] = []
+    log_path = Path.home() / ".hermes" / "logs" / "gui.log"


Use get_hermes_home() / 'logs' / 'gui.log' here. Path.home() / '.hermes' sends logs from an active named profile into the default profile, violating profile isolation.

…update Capture each manually-started dashboard/serve process's argv before the stale-process kill (/proc/<pid>/cmdline on Linux, ps -o command= on macOS), then respawn it detached after the update — headless (--no-open) with output to logs/dashboard-restart.log under the active profile's HERMES_HOME. Supervised PIDs keep their systemd-unit restart; --stop stays a plain stop. Salvaged from PR #41508 with scope fixes: serve matching preserved, profile- aware log path, restart only on the update path (restart_managed=True).

teknium1 · 2026-07-26T20:43:19Z

Merged via PR #72192 — your argv-capture/respawn work was applied onto current main with your authorship preserved in git history (8e1fb9e), with the review-noted adjustments folded in (serve matching kept, profile-aware log path via get_hermes_home(), capture gated to the update path). Thanks for being the first to tackle the manual-restart half.

…update Capture each manually-started dashboard/serve process's argv before the stale-process kill (/proc/<pid>/cmdline on Linux, ps -o command= on macOS), then respawn it detached after the update — headless (--no-open) with output to logs/dashboard-restart.log under the active profile's HERMES_HOME. Supervised PIDs keep their systemd-unit restart; --stop stays a plain stop. Salvaged from PR NousResearch#41508 with scope fixes: serve matching preserved, profile- aware log path, restart only on the update path (restart_managed=True).

alt-glitch added type/bug Something isn't working P3 Low — cosmetic, nice to have comp/cli CLI entry point, hermes_cli/, setup wizard labels Jun 7, 2026

Variable85 added 2 commits June 24, 2026 07:28

fix: restart systemd dashboard services after update

93d997e

Variable85 force-pushed the fix/dashboard-autorestart-after-update branch from 56b75f7 to 93d997e Compare June 24, 2026 07:31

teknium1 reviewed Jul 14, 2026

View reviewed changes

teknium1 mentioned this pull request Jul 26, 2026

fix(update): restart supervised and manually-started backends after update #72192

Merged

teknium1 closed this Jul 26, 2026

Navigation Menu

Uh oh!

fix: auto-restart dashboard after update when argv is recoverable - #41508

fix: auto-restart dashboard after update when argv is recoverable#41508
Variable85 wants to merge 2 commits into
NousResearch:mainfrom
Variable85:fix/dashboard-autorestart-after-update

Variable85 commented Jun 7, 2026 •

edited

Loading

Uh oh!

teknium1 left a comment

Uh oh!

teknium1 Jul 14, 2026

Uh oh!

teknium1 Jul 14, 2026

Uh oh!

teknium1 Jul 14, 2026

Uh oh!

teknium1 commented Jul 26, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

Variable85 commented Jun 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test Plan

Notes

Uh oh!

teknium1 left a comment

Choose a reason for hiding this comment

Problems

Suggested changes

Uh oh!

teknium1 Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

teknium1 Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

teknium1 Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

teknium1 commented Jul 26, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Variable85 commented Jun 7, 2026 •

edited

Loading