Why did the alarm judgment occur before the indicators were fully aggregated?

[SWHHEART]

Why is the alarm judgment carried out before the indicators have been fully aggregated? The position indicators are still in the aggregation process. For example, the value of service_apdex health status during the aggregation process may trigger an alarm, but the final stored value is within the normal threshold range. Does this result in data [inconsistency?]

[wu-sheng]

Why is the alarm judgment carried out before the indicators have been fully aggregated?

How do you get this conclusion? cachedMetrics.combine(metrics) did the aggregation. I feel you are guessing the logic, and not fully understand.

[SWHHEART]

"loadFromStorage(metricsList)" will load data from the database into the cache, or if a certain metric is already in the cache, then in the subsequent logic, it will aggregate the same metric, and then execute "nextWorker" for a trigger verification of the alarm again. Suppose the first set of metrics triggered an alert, but the metrics had not been stored yet. Subsequent metrics of the same type will be aggregated multiple times, and eventually the stored metrics will be the normal ones. Then, can the previous alert be traced back through the data?

[wu-sheng]

I don't know how you make this happen. Cache is entity and time_bucket oriented, they must be aggregated.