Richard Prévot

JedisConnection - Socket not closed properly when redis.quit command fails

1

Hi, When a host is unreachable or read timeouts are encountered, the [JedisConnection.close()](https://github.com/Netflix/dyno/blob/7efd3c36fa7ace3d801e67cbb5c8a7f6f1aa5e2f/dyno-jedis/src/main/java/com/netflix/dyno/jedis/JedisConnectionFactory.java#L164) method doesn't ensure that the socket is properly closed : @Override public void close() { jedisClient.quit(); jedisClient.disconnect();...

Pull Request for issue #292

Health checking for synchronous connection pool has been enabled in PR #112. This mechanism doesn't actually work. In [ConnectionPoolHealthTracker](https://github.com/Netflix/dyno/blob/ad3fa2bf3364e94314371269978ae9370ec5e25f/dyno-core/src/main/java/com/netflix/dyno/connectionpool/impl/health/ConnectionPoolHealthTracker.java#L218), the method pingHostPool is executed periodically by a scheduler: ``` private...

Computes replication factor for complete topology (active and inactive hosts)

2

Pull Request for issue #195

Hi, When using Token Aware load balancing, the replication factor is checked during startup. This validation is done by the HostSelectionWithFallback class (initWithHosts and calculateReplicationFactor methods): ``` public void initWithHosts(Map...

Hi, This pull request fixes #592. Note that msg_get_error() may return a NULL pointer when : 1. Client requests are dropped under heavy traffic (call stack : req_forward_error => msg_get_error...