If that works, it's a neat/fancy trick. But, after looking into the docs
for that setting, related to vnodes which I'm not using, I'm not
comfortable messing with it on my production cluster.
I guess this implies that the # of mappers is related to the number of
physical/virtual nodes of the cassandra cluster, which makes sense...
I think I'm just going to hack up a script that I can run locally on a
cassandra box, and only run queries/writes/deletes against the local token
range (and get parallelism from cassandra's distributed nature).
On Sun, Apr 6, 2014 at 7:30 AM, Suraj Nayak <[EMAIL PROTECTED]> wrote: