[
https://issues.apache.org/jira/browse/TRAFODION-2952?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16588892#comment-16588892
]
liu ming commented on TRAFODION-2952:
-------------------------------------
I propose to add a CQD to control the sequence retry number and time. So for
this kind of special case, there is a way to reduce the sequence error, which
will interrupt the load process.
> large amount of data will cause error in sequence generating
> ------------------------------------------------------------
>
> Key: TRAFODION-2952
> URL: https://issues.apache.org/jira/browse/TRAFODION-2952
> Project: Apache Trafodion
> Issue Type: Bug
> Reporter: liu ming
> Assignee: liu ming
> Priority: Major
>
> insert error
> ** ERROR[1583] Sequence metadata could not be updated.
> create table test1
> (id LARGEINT not null
> ) primary key(id)
> SALT USING 48 PARTITIONS
> ATTRIBUTES ALIGNED FORMAT
> HBASE_OPTIONS
> (
> DATA_BLOCK_ENCODING = 'FAST_DIFF',
> COMPRESSION = 'SNAPPY',
> MEMSTORE_FLUSH_SIZE = '1073741824'
> );
> create table test2
> (id LARGEINT not null
> ) primary key(id)
> SALT USING 48 PARTITIONS
> ATTRIBUTES ALIGNED FORMAT
> HBASE_OPTIONS
> (
> DATA_BLOCK_ENCODING = 'FAST_DIFF',
> COMPRESSION = 'SNAPPY',
> MEMSTORE_FLUSH_SIZE = '1073741824'
> );
> create sequence seq ;
> alter sequence seq cache 20000;
> upsert into test2 select seqnum(seq, next) from test1;
> test1 table has about 4 billion rows. When error raised , next value of seq
> is 5300001
> CREATE SEQUENCE TRAFODION.SEABASE.SEQ
> START WITH 1 /* NEXT AVAILABLE VALUE 5300001 */
> INCREMENT BY 1
> MAXVALUE 9223372036854775806
> MINVALUE 1
> CACHE 20000
> NO CYCLE
> LARGEINT
> ;
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
