could not start gridsql

GridSQL does support serial data type.
You should change
apk_id integer NOT NULL DEFAULT nextval(('pk_aml_id'::text)::regclass),
to
apk_id serial NOT NULL,

Regarding the error, please see the log for the root cause, why node 2 can not send over intermediate results.
Something is wrong with GridSQL or Postgres configuration.
If count() runs against replicated table it is executed on one node, performing load balancing.
Count() against partitioned table runs in 2 steps, first - count() on each node, second - sum() of results of first step.
There is a problem between step 1 and 2, but I am not sure about the cause.

Andrei

when run the select count(*) from the partition table with rows 680,000,
error

Caused by: com.edb.gridsql.exception.XDBMessageMonitorException: Node 2 has aborted execution, cause is: com.edb.gridsql.exception.XDBServerException : Can not send data to Nodes
at com.edb.gridsql.engine.MessageMonitor.checkMessages(Unknown Source)
at com.edb.gridsql.engine.MultinodeExecutor.executeStep(Unknown Source)
... 13 more

but run the select count(*) from the partition table with rows 31

It will not create any error,

hi, Andrei

The problem had been found , I change the gridsql.config file .
my gridsql structure is , two nodes: 172.25.165.179---- node 1 and also the coordinator ; 172.25.165.175 ----node 2

xdb.metadata.database=XDBSYS
xdb.metadata.dbhost=172.25.165.179 --------------- original is 127.0.0.1

xdb.node.1.dbhost=172.25.165.179 ------------------127.0.0.1

xdb.coordinator.host=172.25.165.179 ---------------127.0.0.1
xdb.coordinator.port=6454

all change the 127.0.0.1 to 172.25.165.179 , it is ok now.

hi,

I also want to ask a new question , how did the gridsql work, when copy data to a table that is replicate table stucture, how did it work.

when delete the data from the replicate table , how it work?
ths

hi, Andrei

I want to create a procedure (function ) in the gridsql. it seems not support , or there is a better method for this?

When you inserted the 30 rows, did you use COPY or insert?

Are you using loader defaults in gridsql.config, or did you override any settings?

For (2), was there any other additional information in the log file? Was the problematic node on its own server, or shared with others? Is its profile any different from the other nodes?

Thanks,

Mason