Is this question any different from the one on the thread with the title "Slow Impalad nodes"?
On Mon, Apr 10, 2017 at 2:16 AM, Antoni Ivanov <[email protected]> wrote: > Hi again, > > Any idea how to workaround those type of issues now? > If I understand correctly - Impala is limited by the most under-performing > node > Because fragments get scheduled on "round-robin" principle accross nodes. > Nodes with worse CPU or bad network will slow-down all queries (with > fragments) on them? > > Is that correct? > > And 3rd question. Is there JIRA that I can follow which aims to add > "dynamic scheduling" to Impala. > > Thanks, > Antoni >
