DSXchange

EXECUTE_COMMAND (Uses wildcard to get files and uses the ls -m parameter ......but without a comma ----> StartLoop (this is now looping through a list but looking for a comma which does not exist !" Now why not try the ls -1 too .....you dont have to worry about the comma's... :) . ...but for ...

try this "SELECT COUNT(*) FROM DS_JOBS" in the Datastage administrator...

Yes, also try splitting the job horizontaly if your business rules allow you to do that. Job1 Ds1---- Join 1----- Funnel -------- Sort ------- Ds5 Ds2---- Job2 Ds3---- Join 2----- Funnel -------- Sort ------- Ds 6 Ds4---- Job3 Ds5---- Join 2----- Funnel -------- Sort ------- Ds 7 Ds6----

Also try executing ls -1 which forces the out put into one-entry-per-line and feed it to the start loop activity marking the "List loop" as its loop type.

Did you view the data through the seq file....also check the column ordinalities between the source and the target..

may be develop a server job ....Database server can be taxed for the extraction process....

As the default settings are Parallel (Read) and Propagate(write) in a transformer, this becomes a conflict for the sequential file stage which writes in sequential mode while actually receiving them in parallel. By clearing the default mode, we are paving the way acceptable to the seq file.

Go to the Datastage Administrator and type the following query SELECT CATEGORY FROM DS_JOBS WHERE NAME = 'sequencename';

Try what John has said. I have done it that way before for a similar requirement.....

Pointing the Message file option to the /tmp directory solved the issue.
Looks like the default directory had no write permission.

Hi, Using db2 stage, while an insert is attempted by using the write method 'Load', a fatal error occurs hinting at the following... DB2 Load Binary Error. Error in accessing a file or path of type "TEMP_FILE" during load or load query. Reason code: "1". ....while the other write...

yes! This is exactly what I intend to do.
Thanks.

Well, its a incremental load and the real bottle neck seems to be the sequencer number generated in the database which does not allow the parallel processing in db2. Earlier while generating numbers using a surrogate key stage, the job took 30 secs to run a million records aparently using the luxury...

Well, a sequential load and a collector type of sort merge did raise the performance from 147 rows/sec to 3700 rows/sec. wonder if this could still be increased? A new insert to the table actually showed a 55000 + rows/sec but the database seems to insert at a very low rate. Has anybody got anything...

droping the sequence number generator and placing a surrogate key generator improved the performance to very marginal levels....around 470 rows/sec only