[
https://issues.apache.org/jira/browse/FLINK-20375?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17240581#comment-17240581
]
Mulan Wong commented on FLINK-20375:
------------------------------------
TaskMetricGroup#METRICS_OPERATOR_NAME_MAX_LENGTH truncate operator name, but
task name is too large to lead to Prometheus memory problem.
For example:
task_name: Source: wuren_foo_ods_foo ->
SourceConversion(table=[Unregistered_DataStream_1], fields=[id, name]) ->
SinkConversionToRow ->
SourceConversion(table=[default_catalog.default_database.ods_foo], fields=[id,
name, PROCTIME]) -> Calc(select=[id, name]) -> SinkConversionToTuple2 -> Sink:
Unnamed
> Add a switch to control the operator name in SQL
> ------------------------------------------------
>
> Key: FLINK-20375
> URL: https://issues.apache.org/jira/browse/FLINK-20375
> Project: Flink
> Issue Type: Improvement
> Components: Table SQL / Runtime
> Reporter: hailong wang
> Priority: Major
>
> From the user-zh email:
> [http://apache-flink.147419.n8.nabble.com/FlinkSQL-Prometheus-td8858.html]
> The sql operator name is too large, this resulted in a lot of memory in
> querying Prometheus which is not friendly in the production environment.
>
> I think we can add a switch to control the operator name in SQL.
> A full name helps debug Jobs, and a short name helps to send this as metrics
> which will save bandwidth, CPU consumption during compression, and downstream
> storage.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
