Repository navigation
A fix for schedules that stop firing after a database blip.
- Automations: when the database connection behind a service's advisory locks died (a Postgres restart or failover), every later lock call hung. The automations scheduler then never ran again, so no schedule fired until the service was restarted; the orchestrator's sweeps and chat's room polls shared the same weakness. Lock sessions now reconnect on the next call (#8).
Install or upgrade with curl -fsSL https://ryph.ai/install.sh | sh. Container images are on quay.io/ryph, tagged 0.1.4.