xiaohui-sun commented on a change in pull request #5142: [TE] Detection
creation endpoints should trigger Replay & Tuning
URL: https://github.com/apache/incubator-pinot/pull/5142#discussion_r391408969
##########
File path:
thirdeye/thirdeye-pinot/src/main/java/org/apache/pinot/thirdeye/detection/yaml/YamlResource.java
##########
@@ -380,6 +380,9 @@ long createDetectionConfig(@NotNull String payload, long
startTime, long endTime
Long id = this.detectionConfigDAO.save(detectionConfig);
Preconditions.checkNotNull(id, "Error while saving the detection
pipeline");
+ // create an yaml onboarding task to run replay and tuning
+ createYamlOnboardingTask(detectionConfig.getId(), startTime, endTime);
Review comment:
> The onboarding jobs complete within half a minute on an average case and I
don't think it should be a problem. In addition, I am not sure if
bulk-onboarding of alerts is a common use-case or whether we should even
encourage such use-cases except for some special cases like "onboard all UMP
metrics" in which case we should anyways take precaution. Alerts are typically
created by users on various portals and the throughput(requests/sec) is
typically not that high.
>
> On the other hand, if we do not trigger the onboarding job, then apart
from the replay, even the tuning component isn't triggered.
Definitely this is not encouraged. We need to have a fuse mechanism
somewhere to avoid bringing down the whole site if it happens. Maybe we should
throttle the replay tasks.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
