- Reduce authorization retry attempts from 3 to 2 and optimize retry delays from exponential (3s, 5s, 8s, 12s) to fixed 3s backoff
- Implement team-level circuit breaker: skip member processing when 3+ consecutive 500 errors detected in same team
- Add consecutive 500 error tracking with atomic counter and reset on successful authorization
- Reduce ObtainAuthorizationCode timeout from 3 minutes to 90 seconds with updated error messages
- Optimize Codex API workspace selection retry: reduce max attempts from 5 to 3 with shorter backoff (2s, 4s instead of 3s, 5s, 8s, 12s)
- Reduce S2A OAuth submission retry delays from (4s, 6s) to (2s, 3s) for faster failure detection
- Optimize member stagger timing: reduce from 3s + idx*2s to 1s + idx*1s with reduced jitter (0-1s instead of 0-2s)
- Add early exit for exhausted retries in CompleteWithCodexAPI to prevent unnecessary outer retry attempts
- These changes improve responsiveness and reduce cascading failures during bulk team processing
- Add exponential backoff with jitter to S2A OAuth submission (3 attempts)
- Implement 5xx error retry mechanism in Codex API workspace selection (5 attempts)
- Add 429 rate limit handling with retry support in Codex API
- Improve team member processing with staggered delays to avoid concurrent conflicts
- Add per-attempt proxy rotation to avoid reusing failed proxies
- Enhance retry delay calculation with random jitter to prevent thundering herd
- Update logging to display retry attempts and delay durations
- Improve error messages with HTTP status codes and response body snippets
- Refactor retry loops to use consistent exponential backoff pattern across modules
- Add CleanupStuckProcessingOwners database method to reset owners stuck in processing status back to valid state
- Call cleanup routine on application startup to recover from previous abnormal exits
- Call cleanup routine after batch processing completes to ensure no orphaned processing states remain
- Add markOwnerResult calls in processSingleTeam when team invitations are full to properly track owner status
- Prevents owners from being permanently stuck in processing state due to unexpected shutdowns or errors
- Implement pause/resume mechanism using sync.Cond to halt worker goroutines
- Add consecutive zero S2A counter to detect when no accounts are added for 2+ teams
- Trigger automatic 1-minute pause when consecutive zero entries detected
- Log warning messages with team details when pause is triggered
- Reset counter on successful S2A entries or when pause completes
- Allow stop signal to interrupt pause period for graceful shutdown
- Broadcast resume signal to all waiting workers after pause completes
- Prevents resource exhaustion and provides recovery time during low-activity periods
