Monitor and manage services

Day-to-day operations take a few commands: check health, read logs, and apply configuration changes. All of them are safe to run on a live installation.

Check health

helpin doctor --dir /srv/helpin

Doctor checks Docker, your Compose configuration, required services, API readiness, public HTTPS, secret file permissions, and product capabilities. For each problem, it prints the next step.

In the app, server admins see the same checks under Settings → System status. Workspace members can query GET /api/workspaces/{id}/capabilities.

See what's running

helpin status --dir /srv/helpin

Status lists every container, its image identity, and the current migration head. Long-running services should be running (healthy). The one-shot jobs (helpin-migrate, temporal-schema, temporal-namespace) should show exited (0).

Read logs

helpin logs --dir /srv/helpin                       # all services
helpin logs --dir /srv/helpin helpin-api            # one service
helpin logs --dir /srv/helpin agent-runtime agent-runtime-worker

Symptom	Start with these services
Dashboard or login problems	`helpin-api`, `helpin-migrate`, `helpin-frontend`
Live chat not updating	`helpin-api`, `nats`, `redis`
Agents not responding	`agent-runtime`, `agent-runtime-worker`, `helpin-api`
Indexing or scheduled jobs	`helpin-worker`, `temporal`
Attachments	`garage`, `helpin-api`
Public help center	`helpin-helpcenter`

Privacy: Logs can contain customer data. Review excerpts before you share them, and never share .env, signed attachment URLs, or backups.

Apply configuration changes

After you change .env or run helpin configure, apply the change:

helpin restart --dir /srv/helpin

Restart recreates services and keeps all data. Settings that server admins change in the app, such as application email or the signup policy, apply immediately without a restart.

Stop and start

helpin stop --dir /srv/helpin
helpin start --dir /srv/helpin

stop removes containers and networks but keeps your data volumes.

Warning: Never run docker compose down -v on an installation you want to keep. The -v flag deletes every data volume.

Capacity

The evaluation baseline is 8 GB RAM and 20 GiB of free disk. Watch these as usage grows:

Disk: PostgreSQL and Garage grow with conversations, documents, and attachments. Docker's own data disk also needs room for images.
Agent workload: chat, agent runs, and code execution share one agent-runtime-worker. Heavy runs compete with interactive chat, so size CPU and memory for your busiest agent hours.
Rate limits: tune AUTHENTICATED_RATE_LIMIT_PER_MINUTE and EXPENSIVE_RATE_LIMIT_PER_MINUTE if legitimate automation hits 429 responses.

Was this article helpful?