apache storm reliablity timeout configuration -
i have nodejs->kafka>storm->mongo deployed in linux ubuntu. normal originally. changed method in storm worker makes storm worker process message slow, around 1 minute per message, notice message sent again , again storm. revert original method, fine. (original method process time 90ms per message).
i guess storm reliability come player. when message not acknowledged, or time out, sends message again.
if guess right, how configure timeout?
if guess wrong, why same message sent twice or 3 times?
you can set timeout via configuration parameter config.topology_message_timeout_secs
. see https://storm.apache.org/javadoc/apidocs/backtype/storm/config.html#topology_message_timeout_secs
the default value 30 seconds, see defaults.yaml
here: https://github.com/apache/storm/blob/master/conf/defaults.yaml
# maximum amount of time message has complete before it's considered failed topology.message.timeout.secs: 30
when tuple fails, should show in storm ui , should logged, (maybe need adjust log level). can double check if tuple times out or not.
Comments
Post a Comment