Same query performance for 2 historical nodes as for 3 historical nodes

Hi guys,

I had 3 histo nodes and a query ran in ~17-18 seconds.

After that i’ve shut down 1 of them and the query ran in the same time.

I tried to comment the cache enabling lines in broker’s config and restarted the broker, thinking that it’s because of broker query cache:
#druid.broker.cache.useCache=true
#druid.broker.cache.populateCache=true

but the result was the same.

Any idea what i’m doing wrong?

Thanks, Vadim. Vararu.

What query are you running? 17-18 seconds seems extremely slow, but Druid performance is very finicky around tuning.

A groupBy on 2 dimensions + 1 count over all data

To be mentioned that we store raw data. Query granularity - ALL.

Vadim,
Can you post the configuration of the hardware along with the druid configurations?

Did you try to find if any of the historical node was bottlenecked on cpu or memory??

Thanks

Rohit