Hi Vincent,


The table creation statement is similar to below. We have about 200 fields. Table is mutable and don’t have any index on the table.



      state CHAR(2) NOT NULL,

      city VARCHAR,

      population BIGINT,


      CONSTRAINT my_pk PRIMARY KEY (state));





On Wed, Dec 12, 2018, 13:42 Vincent Poon <vincentpoon@apache.org wrote:
For #2, can you provide the table definition and the statement used?  e.g. Is the table immutable, or does it have indexes?

On Tue, Dec 11, 2018 at 6:08 PM Shawn/Xiang Li <shawnlijob@gmail.com> wrote:



1.       Want to check what is underlying running for limit clause used in the following Upsert statement (is it involving any coprocessor working behind?):


                                  upsert into table2 select * from table1 limit 3000000;  (table 1 and table 2 have same schema)


              The above statement is running a lot slower than without “limit”  clause as shown in following, even the above statement upsert less data:


                                upsert into table2 select * from table1;


2.       We also observe memory usable is pretty high without the limit clause (8gb vs 2gb), sometimes for big table it can reach 20gb without using limit clause.  According to phoenix website description for upsert select “If auto commit is on, and both a) the target table matches the source table, and b) the select performs no aggregation, then the population of the target table will be done completely on the server-side (with constraint violations logged, but otherwise ignored).”


               My question is If everything is done on server-side, how come we have such high memory usage on the client machine?