- queryWithRetry now uses explicit client checkout; on transient error,
calls client.release(true) to DESTROY the dead connection instead of
returning it to pool. Fresh connections are created on retry.
- connectWithRetry validates connections with SELECT 1 before returning
- Health check destroys bad connections on failure
- Reduced idleTimeoutMillis from 30s to 10s for faster stale connection eviction
- Fixes BUG-075: pool kept reusing dead TCP sockets after PgBouncer pod restart
- Enable TCP keepalive on pg.Pool to detect dead connections
- Add connectionTimeoutMillis (5s) to prevent hanging on stale connections
- Add queryWithRetry() with exponential backoff for transient DB errors
- Add connectWithRetry() for transaction-based operations
- Detect PgBouncer "no available server" and other transient errors
- Health check has 3s timeout and returns 503 on DB failure
- All DB operations in keys, verification, usage use retry logic
Fixes BUG-075: PgBouncer failover causes permanent pod failures
- Launch BROWSER_COUNT separate Chromium instances (default: 2)
- Each with PAGES_PER_BROWSER pages (default: 8, 16 total)
- Round-robin distribution across browser instances
- Independent restart scheduling per browser
- Updated health endpoint to show per-browser stats
- docker-compose: added BROWSER_COUNT and PAGES_PER_BROWSER env vars
