The problem
A job scheduled for Fridays at 04:00 that fails at 04:00 waits until next Friday. There is no retry, so a transient failure — a database that was briefly unavailable, a network blip — costs a full cycle.
What we have in mind
Retry executions configurable per process, along the lines of what App Engine's cron retry parameters cover:
- A maximum number of retry attempts
- A time limit measured from the original scheduled time
- Backoff between attempts
Both bounds applying together, so a job stops retrying when either is reached.
Opened from our own backlog when this tracker went public. If you need this, say so — comments and concrete use cases are what move something up the list.
The problem
A job scheduled for Fridays at 04:00 that fails at 04:00 waits until next Friday. There is no retry, so a transient failure — a database that was briefly unavailable, a network blip — costs a full cycle.
What we have in mind
Retry executions configurable per process, along the lines of what App Engine's cron retry parameters cover:
Both bounds applying together, so a job stops retrying when either is reached.
Opened from our own backlog when this tracker went public. If you need this, say so — comments and concrete use cases are what move something up the list.