Revise container failure handling to use error state instead of cancelled
- Dominant language
- Python
- Stars
- 670
- Forks
- 183
- Avg merge
- 17h 7m
- Merged PRs (30d)
- 358
Description
## Current Behavior
When container creation fails, the session transitions to `CANCELLED` state, stopping all operations without retry attempts.
## Proposed Change
Use `ERROR` state for container creation failures instead of `CANCELLED`.Reserve `CANCELLED` state only for:
- Explicit destroy requests before container creation
- Failures during preparation phase
## Rationale
1. In multi-container sessions, container creation failures should allow cleanup:\* Current: Manual intervention needed to clean up successful containers
- Proposed: Use `destroy` functionality on `ERROR` state for cleanup
1. Aligns with k8s pattern of failed state handling with retry logic
1. Determine the required API changes and assess their impact
## Implementation Suggestion
Flow: Creation failure → `ERROR` state → retry attempts → if all attempts fail → final error handling
## Impact
Requires state transition logic changes
JIRA Issue: BA-55
Contributor guide
Assessment
This issue has not been assessed yet.