alnzng commented on code in PR #1615:
URL: https://github.com/apache/samza/pull/1615#discussion_r909099191
##########
samza-core/src/main/java/org/apache/samza/startpoint/StartpointManager.java:
##########
@@ -317,15 +317,55 @@ public Map<TaskName, Map<SystemStreamPartition,
Startpoint>> fanOut(Map<TaskName
* @return fanned out Startpoints
*/
public Map<SystemStreamPartition, Startpoint> getFanOutForTask(TaskName
taskName) throws IOException {
+ return getStartpointFanOutPerTask(taskName)
Review Comment:
In general, metrics make more sense than logging if the monitoring
behaviors/operations are producing time-series data, which means the data are
kept updated frequently in the app's lifecycle. Either metrics and logs can
have their own TTL, which depends on the SLA for their backend storage
services. Also, one con for metrics is that they can't include too much
information for troubleshooting because of the high-cardinality challenge in
the metrics storage.
For this particular use case "startpoints", it was intended for once in life
and all its related operations are usually only invoked very few times in the
very early phase of the app's lifecycle. I would prefer to keep using logs than
metrics since we can add as much as information we want to help troubleshoot.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
