If you want to guarantee that the data is written to all nodes before the code returns, then yes you have to use "consistency all". Otherwise there is a small risk of outdated data being served if a node goes offline longer than hints timeouts.
Somewhat looser options that can assure multiple copies are written, as you probably know, are quorum or a hard coded value. This applies to a typical installation with a substantial number of nodes of course, not a small 2-3 node cluster. I am curious why localStrategy when you have such concerns about data consistency that you want to assure all nodes get data written. Can you elaborate on your use case? *.......* *“Life should not be a journey to the grave with the intention of arriving safely in apretty and well preserved body, but rather to skid in broadside in a cloud of smoke,thoroughly used up, totally worn out, and loudly proclaiming “Wow! What a Ride!” - Hunter ThompsonDaemeon C.M. ReiydelleUSA (+1) 415.501.0198London (+44) (0) 20 8144 9872* On Sun, Mar 15, 2015 at 8:11 PM, 鄢来琼 <laiqiong....@gtafe.com> wrote: > Could you tell me whether the meta data of the new table are build in > all the nodes after execute the following statement. > > > > *cassandra_session.execute_async(* > > *“““**CREATE TABLE Table_test(* > > * ID uuid,* > > * Time timestamp,* > > * Value double,* > > * Date timestamp,* > > * PRIMARY KEY ((ID,Date), Time)* > > *) WITH COMPACT STORAGE;* > > *”””* > > *)* > > > > As I know, the system keyspace is used to store the meta data, but the > strategy is localStrategy, which only store meta data of local node. > > So I want to know whether table is created in all the nodes, should I add > consistency_level setting to the above statement to make sure > > “create table” will be executed in all the nodes? > > Thanks. > > > > Peter > > >
