1) correct
2) is the number of measurements; due to the random distribution of
latency markers this value can be surprisingly low depending on the
latency marker interval
3) I don't know, but it isn't exposed by Flink.
On 08.10.2018 10:17, varuy322 wrote:
Hi there,
I have integrated kafka with flink1.5, also with the help of prometheus and
Granada to display metrics of flink 1.5,
now i get three indicator about latency as below:
1)flink_taskmanager_job_latency_source_id_source_subtask_index_operator_id_operator_subtask_index_latency(short
for latency)
2)flink_taskmanager_job_latency_source_id_source_subtask_index_operator_id_operator_subtask_index_latency_count(short
for latency_count)
3)flink_taskmanager_job_latency_source_id_source_subtask_index_operator_id_operator_subtask_index_latency_sum(short
for latency_sum)
it seems 1) to be the latency value of one real subtask,
for 2),latency_count always equal to one const, in my example the value is
128,
for 3),latency_sum always equal to zero.
could you tell me what's the meaning of 2) and 3)?
Best regards & Thanks.
Rui, Wang
<http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/file/t1661/latency_metric.png>
-----
stay hungry, stay foolish.
--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/
