The Vertica Database Forums

I have a query that has sporadic run times. When viewing the execution_engine_profiles on one of the long runs, it is spending over 300 seconds on "output queue wait(us)" from operator "NewEENode". Vertica documentation definition is "Time in microseconds that an operator spends waiting for the output buffer to be consumed by a downstream operator." How do I know what this means or what operator it is waiting on?

Hi Janica,

There is probably some other operator that has a long "clock time (us)" counter_value in the execution_engine_profiles data. You can correlate the profiling data to the explain plan using the path ids.

--Sharon

In my case , i can see high amounts of network wait (us) in the execution_engine_profiles such as:

v_padb_node0001 | NetworkRecv | 1 | network wait (us) | Net id 1001 - v_padb_node0003 | 20612175
v_padb_node0001 | NetworkRecv | 1 | network wait (us) | Net id 1001 - v_padb_node0002 | 20663181
v_padb_node0001 | NetworkRecv | 1 | network wait (us) | Net id 1001 - v_padb_node0001 | 0

Like for a query taking 30-35 seconds , the network wait is about 20-23 seconds, so does that mean that there is some network latency in between the traffic between the nodes. Or is there any issue at the system side.