Hi team, I am using

Volume_anomalies

for an incremental table. Sometimes, we need to update that model, so we need to drop it and re-populate it, which causes an anomaly. (moving from any size to zero and then to any size again). eventual drops are expected, how can we set this test to ignore those events and prevent failures? Is there any sensitivity parameter that we can set? I really appreciate any help you can provide.

Hi @christoph Nang. Do you know the timing of when these full updates are happening? If so you are right and

anomaly_exclude_metrics

is an option. In the cloud product it would learn these patterns, but that is not something that is supported in the open source unfortunately (as it requires some model training).