A question about connectionTimeout and connection

[RonaldWakeup]

Hello, I Use curator as follow:
CuratorFrameworkFactory.Builder builder = CuratorFrameworkFactory.builder()
.connectString("xxx") //this xxx includes 5 invalid IP address.
.sessionTimeoutMs(10_000)
.connectionTimeoutMs(10_000)
retryPolicy(retryNtimes(3,5000));
client.start();
I try to test curator's timeout action when curator first attempt to connect zk server. I think curator(actually internal zookeeper client) will not try to connect server after it retry 3 times and connectionTimeout，but actually I see the log file，zk client seems not controlled by curator? the zk client always throw some exception, such as connectionLossException... never stop until I stop my program.
I use curator not long ago, and try to read curator code. But I still can't understand why curator behavior as described above.
Hope to receive your reply soon.
Thank you.

[kezhuw]

I think curator(actually internal zookeeper client) will not try to connect server after it retry 3 times and connectionTimeout，but actually I see the log file，zk client seems not controlled by curator?

RetryPolicy has nothing to do with start, it applies only to zookeeper data operations, say getData, setData and so on.
You probably should use Watcher in zookeeper or ConnectionStateListener in curator to receive state change events.

zk client seems not controlled by curator? the zk client always throw some exception, such as connectionLossException... never stop

It is zookeeper's internal behavior. See also ZOOKEEPER-4508, ZOOKEEPER-4921 and ZOOKEEPER-4923

until I stop my program.

Does client.stop help ?