nodetool repair -pr on every node --- covered all ranges of data. On Sun, Jul 1, 2018 at 7:03 AM, Riccardo Ferrari <ferra...@gmail.com> wrote:
> Hi list, > > After long time of operation we come to the need of growing our cluster. > This cluster was born on 2.X and almos 2 years ago migrated to 3.0.6 ( I > know we are bit prudent ) > > The cluster was a 3 m1.xlarge (we are on AWS) and table RF was 3 > > Thanks to your valuable hints we added a new node in the same AZ that > joined flawlessy after making sure the streaming_socket_timeout_in_ms was > high enough. > > I was used to run: > > - nodetool repair -pr (on each node and alterante days) > > QUESTION TIME: > > Q1: Should I keep it that way or should I run some "nodetool repair -full" > on all the nodes the first time ? > > Reading today's documentation (http://cassandra.apache.org/ > doc/latest/operating/repair.html) it's not really clear to me what should > be the best practice. Any pointer is much appreciated > > Thanks, >