Adjusting Heartbeat TCP Timeout Interval
Using the TCP flag, you can prevent the Statestore from waiting indefinitely for a response from the subscribers that fail to respond to the heartbeat RPC within the set period.
This flag statestore_heartbeat_tcp_timeout_seconds
defines the time
that may elapse before a heartbeat RPC connection request from a Statestore
server to an Impalad or a Catalog server (subscribers) should be considered dead.
You can increase the flag value if you see intermittent heartbeat RPC timeouts listed in the statestore's log. You may find the max value of "statestore.priority-topic-update-durations" metric on the statestore to get an idea of a reasonable value to be used in this flag.
- In Cloudera Manager, navigate to .
-
In the Impala Daemon Command Line Argument Advanced Configuration
Snippet (Safety Valve) field, add the flag
statestore_heartbeat_tcp_timeout_seconds
with an appropriate value. -
You can also control the maximum number of consecutive heartbeat messages an
impalad can miss before being declared failed by the statestore by adding this
flag
statestore_max_missed_heartbeats
. Typically, you will not have to change this value. - Click Save Changes and restart Impala.