Neoview Database Administrator's Guide (R2.2)

Example 8-4 (page 131)shows the task instance log file for a extract task that completed without
errors:
Stopping an Extract Task
To abruptly stop an extract task while it is running, enter a gcmd command on the Linux platform
with requestType STOP_TASK:
gcmd -rt STOP_TASK argument-list
For example, to stop the task started in Step 2:
gcmd -rt START_TASK -tk DAILY-EXTRACT -ti OCT01 -un role.dba -pw roledba
For descriptions of each command argument, see “Arguments for Stopping a Task ” (page 114).
Failed Extract Tasks
If the Linux or Neoview gelserver extract process fails, the extract task is aborted. The task
instance log reports that the extract completed with errors.
Remove any partial output files, then rerun the Extract task. If the output file already exists it
will be renamed and the Extract task starts over from the beginning. There is no Recovery feature
for extract tasks.
Extract Task Modes
Extract tasks have two modes:
• Parallel mode
This mode is used when you extract data from a single table. In this mode, multiple Extractor
gelserver processes are started on each Neoview segment, where there are partitions for the
source table. Each Extractor process on the Neoview system extracts the data from a single
partition of the source table. When executing the SELECT query, an Extractor process restricts
the SELECT to a single partition by using the LOCATION clause to modify the query you
provide. For example, suppose the extract query is:
SELECT * FROM NEO.DB.T1
The Extractor process handling the data for the partition on segment \NEO0001, disc $FC1200,
would modify the query to:
SELECT * FROM TABLE(TABLE NEO.DB.T1, LOCATION
\NEO0001.$FC1200.ZSDXXXXX.XXXXXXXX)
The same technique can also be applied to JOINs where the table specified as the partitioning
table will be modified in the query to restrict the query by performing a join on a single
partition of the table. For example, suppose the extract query is:
SELECT *FROM T1, T2 WHERE T1.C1 = T2.C2
The Extractor process handling the data for the partition on segment \NEO0001, disc $FC1200,
would modify the query to:
SELECT * FROM TABLE(TABLE NEO.DB.T1,
LOCATION \NEO0001.$FC1200.ZSDXXXXX.XXXXXXXX),
T2 WHERE T1.C1 = T2.C2
On the Linux platform, a single multi-threaded Extractor is responsible for receiving the
data from the Neoview platform and writing to the specified output files or pipes. In parallel
mode, there will be one Reader thread allocated to each Extractor process running on the
Neoview platform, and one Writer thread allocated to each output file or pipe.
When the query is partitioned, as described above, you must take care when running extract
queries in parallel mode to ensure that the results from the query are correct. Due to the
complexities of queries, it is not possible for the Extractor process to validate whether a
Extracting Data 123