I'm trying to run a user defined aggregate query to periodically roll up some values in a large table. I'm only trying to aggregate one partition at a time, but each partition can have lots of rows.
Once I get past a certain number of rows in a partition, the query starts to fail with a timeout error since it seems to be using a fixed timeout value.
But I'm in no hurry for the aggregate query to return results, and I'd like the coordinator node to take its time and aggregate all the rows without timing out. I'd also like to throttle the rate at which it pages through the results so that it doesn't generate excessive load while doing the aggregation.
So is there a way to set a timeout and throttling rate for queries that are expected to take a long time to execute?
Right now I'm thinking the only workaround is to design my schema so that I keep my partitions artificially small to keep the aggregation under the fixed timeout limit.
I'm running the Cassandra 3.0 alpha code base.
