Abhishek Agarwal

@paul-rogers - I have added more explanation now in the PR. can you review again?

@tanisdlj - can you share the overlord logs with us?

the build has likely failed due to a code coverage check.

thank you for explaining @tejaswini-imply. I understand it better now and I am at more ease 😄 can you look into the build failures? Would like to see a green...

@FrankChen021 - have you run any performance tests/benchmark for ingestion time rollup? That will be handy to rule out any perf bug.

@FrankChen021 - This is a useful capability. would you be able to take this to completion? Perf tests will be nice but they need not block this PR.

Awesome. with this capability, we can do shuffle joins, window aggregations, etc without bottlenecking on the broker. have you also thought about elasticity? Right now, scaling historicals up and down...

is this change (https://github.com/apache/druid/pull/12167) by any chance? We were not handling the pausing tasks really well.

The ability to cancel the query via console is already available (https://github.com/apache/druid/pull/11738). BTW how long do these queries run typically run?

The router broadcasts DELETE requests to all the brokers. So you can send the cancel request to any router. The console does it the same way.