You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
cron: automatic owner failover when the owner is unavailable #117
Cron ownership only moves when the CLI reconciles (after deploy, restart, rollback, or scale). If the owner's host goes down, its jobs stop running until someone redeploys. Specs are agent-local and never replicated, so no other agent knows to take over. This is a feature addition: the website documents "Owner failover is not automatic."
Scope
Automatic owner failover when the current owner is unavailable.
Needs a design for who decides (agents via the replicated catalog vs. the CLI), how a new owner gets the spec without replication, and how double runs are avoided when the old owner comes back (fencing).
Done when
The cron reference on the website ("Ownership" and "Current Limits") and AGENTS.md are updated in the same change.
Activity