mindesbunister 8717f72a54 fix: Add retry logic to exit order placement (TP/SL) ...

CRITICAL FIX: Exit orders failed without retry on 429 rate limits

Root Cause:
- placeExitOrders() placed TP1/TP2/SL orders directly without retry wrapper
- cancelAllOrders() HAD retry logic (8s → 16s → 32s progression)
- Rate limit errors during exit order placement = unprotected positions
- If container crashes after opening, no TP/SL orders on-chain

Fix Applied:
- Wrapped ALL order placements in retryWithBackoff():
  * TP1 limit order (line ~310)
  * TP2 limit order (line ~334)
  * Soft stop trigger-limit (dual stop system)
  * Hard stop trigger-market (dual stop system)
  * Single stop trigger-limit
  * Single stop trigger-market (default)

Retry Behavior:
- Base delay: 8 seconds (was 5s, increased Nov 14)
- Progression: 8s → 16s → 32s (max 3 retries)
- Logs rate_limit_recovered to database on success
- Logs rate_limit_exhausted on max retries exceeded

Impact:
- Exit orders now retry up to 3x on 429 errors (56 seconds total wait)
- Positions protected even during RPC rate limit spikes
- Reduces need for immediate Helius upgrade
- Database analytics track retry success/failure

Files: lib/drift/orders.ts (6 placePerpOrder calls wrapped)

Note: cancelAllOrders() already had retry logic - this completes coverage

2025-11-15 17:34:01 +01:00

32 KiB

Raw Blame History

View Raw