alertlogic/erllambda

need to flush out std out before invoke next

motobob opened this issue · 2 comments

example of spill logs between invokes

MONITORING|1542361428|31|histogram|search.sort.batch.time|#operation:map_write
MONITORING|1542361428|32|histogram|search.sort.batch.time|#operation:create_new_map
MONITORING|1542361428|1292|histogram|search.sort.invoke.time|#invoke_type:merge_batches
END RequestId: 1d6b832c-e984-11e8-a0c1-51c269721f59
REPORT RequestId: 1d6b832c-e984-11e8-a0c1-51c269721f59	Duration: 1297.37 ms	Billed Duration: 1300 ms Memory Size: 2048 MB	Max Memory Used: 746 MB	
START RequestId: 1e375de9-e984-11e8-b495-17f47950124b Version: $LATEST
MONITORING|1542361428|5|count|search.sort.lhttpc|#conn,end
Invoke Success path 1542361428640 http://127.0.0.1:9001/2018-06-01/runtime/invocation/1d6b832c-e984-11e8-a0c1-51c269721f59/response
Invoke Next path 1542361428641 http://127.0.0.1:9001/2018-06-01/runtime/invocation/next

notice success log of previous invoke printed after new one actually started.

@velimir0xff

we might need to wait until the logs are flushed

not an issue after recent changes