Yaliang Wu

Results 253 comments of Yaliang Wu

@dreamer-89 Sure, you can take it. Thanks a lot.

Seems duplicate issue with this one, https://github.com/opendistro-for-elasticsearch/alerting/issues/27

Thanks @sudiptoguha for the detailed explanation. Seems it's not reasonable to calculate every point's nearest centroid inside RCF summary. Is it possible to provide some utility API in RCF to...

There is duplicate issue https://github.com/opendistro-for-elasticsearch/anomaly-detection/issues/217

Same for [ThresholdResultTransportAction](https://github.com/opendistro-for-elasticsearch/anomaly-detection/blob/1637128473cea66db96b87b20a4c3ff7eee5a150/src/main/java/com/amazon/opendistroforelasticsearch/ad/transport/ThresholdResultTransportAction.java#L45) My local log shows `[2020-04-07T10:20:18,579][ERROR][c.a.o.a.t.ThresholdResultTransportAction] [4c327594166b.ant.amazon.com] ElasticsearchTimeoutException[Cannot get response within time limit: get [.opendistro-anomaly-checkpoints][_doc][izSiQXEBZ8BuTDHZ70NC_model_threshold]: routing [null]]`

[ADStateManager](https://github.com/opendistro-for-elasticsearch/anomaly-detection/blob/development/src/main/java/com/amazon/opendistroforelasticsearch/ad/transport/ADStateManager.java#L131) `ElasticsearchTimeoutException[Cannot get response within time limit: get [.opendistro-anomaly-detectors][_doc][WQXnUXEB44dtcje-PgWE]: routing [null]]`

And we should not show create alert button if user doesn't have create monitor permission, or disable the button.

## Root cause Detector first initialization done, then we store model checkpoints to checkpoint index. Then AD plugin disabled, detector stopped automatically without deleting model checkpoint. Then remove 1 feature...

Yes, when user stop detector, AD will delete checkpoint automatically. But when we stop detector for some exception, we don't delete checkpoint. I'm working on job runner to add checkpoint...

Yes, agree. We can't trust the case for dimension equals feature count. At least we can check the not equal case. Ideally, we can store detector configuration in model checkpoint...