@User This is not documented yet. Here are the steps: 1. Go to http://localhost:1426/thirdeye-admin 2. Go to metrics tab 3. Select the data set 4. Click add a new record 5. Fill in the name for the derived metric 6. Set active to true 7. Set derived to true 8. Filling in the expression for the derived metric. Operators should be in “+ - * /”. Metrics should be written as id[metricId]. For example: id2/id1 9. Click save

Example for adding a derived metric

Exactly what I needed, thanks. One more thing, create alert won't let me add a subscription group as I don't have a defined application. I can't find it in code and [sandbox] from docs doesn't work either. Am i missing something should I just manually add it into database? I belive I would need to add it both to entity and application tables

Yes. Application has to be defined via separate api

I didn't find an api for it, got it to work by manually inserting it into database, thanks

Hi all, we are currently working on a new feature called

Controller Separation between Helix and Pinot

. Here is the design doc: https://cwiki.apache.org/confluence/display/PINOT/Controller+Separation+between+Helix+and+Pinot Please feel free to review it and make any comments on it. Thanks!

Hi I have come across an use-case where I need to get multiple aggregated values against a single identified for eg: i have a schema with id, player_name, match_number as a dimension and runs_scored and wickets_taken as metric. I want to find out for a top two players by sum of runs and the no of total wickets taken by them, i.e. sum(runs_scored) and sum(wickets_taken). So my query would be select player_name, sum(runs_scored), sum(wickets_taken) from player_data(table name for eg.) group by player_name order by sum(runs_scored) top 2. But this query is returning two different result set of aggregated values. One aggregation set is having player_name as A and B with runs_scored as 50, 70 respectively and the other aggregation is having name as B and C with wickets_taken as 3, 4 respectively. But i was expecting either A and B or B and C as a result set with runs_scored and wickets_taken metric. Can someone please explain whether this is possible in Pinot or not if yes can someone explain how we can achieve the same?

@User This is the current expected (non-SQL) behavior in Pinot, where orderby is not really supported (ie each aggreation function is always ordered by its own value). One workaround is to get the top to players by

runs_scored

in first query, and then pass the top two player names as filter in the second query so the result is guaranteed to have the two players in the second query's result.

Rooting hard for the India team to win in

CWC-2019

👍

Thanks for update @User 🙂

@User: are we going to add this capability in future ie. multiple aggregation on same identifier ? any idea?

@User yes there are discussions around it, but not timeline yet

OK....thanks for Update @User....Will i be getting update on this channel for the same?

Sure

<!here> I am new to here. is there a slack channel for thirdeye? I am having a problem with 'Quick Guide' with the latest code and would like to ask a question.

ty

@User is it possible to send me an invitation to tysaito@gmail.com for apache-thirdeye.slack.com?

sent it

got it

diff for adding config to support metrics prefix customization: https://github.com/apache/incubator-pinot/pull/4392 appreciate if someone can review it sometime

@User can you also send me an invite for Thirdeye slack at vasilije.markovic@omio.com

@User you can self register to Pinot slack community using https://communityinviter.com/apps/apache-pinot/apache-pinot

Will send one for ThirdEye as well

@User https://join.slack.com/t/apache-thirdeye/shared_invite/enQtNjkxNzUwMzU3NjAwLTUzODgyZTcwMmQwNTU2ZTc5YjNjOGE0MzJmYTdlYmQ1ZTJkNjg1MTY5OWNhY2U4MzZlMWE2MjU1ODliY2JiNGI

I've updated the readme to include the community inviter for both Pinot & ThirdEye to automate the new user on-boarding to Slack. https://github.com/apache/incubator-pinot/pull/4424

thanks

@User @User appreciate if some of you can review this diff when you guys have time: https://github.com/apache/incubator-pinot/pull/4392/files

Hi, at my company we have a mysql table storing per database - table - column - day a variety of summary metrics (countnull, approx count distinct, min, max, p50 etc). Can I use ThirdEye purely for anomaly detection and alerting based on these preaggregated metrics?

Secondly, I followed https://thirdeye.readthedocs.io/en/latest/import_sql_metric.html#import-sql-metric and saw the database popping up in the http://localhost:1426/app/#/self-serve/import-sql-metric dropdrown. But when finalizing the metric, nothing shows up at the 'root cause analysis' tab. What debugging options do I have to figure out the issue here?