Failing task due to incomplete data

[AakarSharmaHME]

Sometimes Anomaly detection job fails or the RCA job fails due to fewer data in a few buckets in that case we don't even show the results for the other buckets.

    if not allTasksSucceeded:
        runStatusObj.status = ANOMALY_DETECTION_ERROR

We have added such checks to the code. But I think this is not a good thing to do as the user is not aware of the reason for the failure. Better is to skip this bucket and show the result of the rest of the buckets.

[AakarSharmaHME]

@vincue Let me know if you agree with my point, I will send a PR to fix the above issue.

[vincue]

Correct me if I'm wrong @AakarSharmaHME , by bucket you mean Anomaly Definitions? The keyword allTasksSucceeded here means all tasks within an anomaly definition, so that wouldn't show error for other Anomaly Definitions if that's what you meant.

[AakarSharmaHME]

By bucket, I mean the values of dimension under consideration for an Anomaly Definition. This is a snapshot of code that I took from anomalyDetectionTasks.py. Here allTasksSucceeded is defined as
allTasksSucceeded = all([anomalyTask["success"] for anomalyTask in result]).

So for example, if a user has selected the Top 10 Dimension value and the anomaly detection task is successful only for 8 of them then we should show anomaly detection of the 8 instead of marking the status as ERROR.