direct path error Qlik Replicate

[trioazhari]

Dear All,

I have a task to full reload 1 database, there are some tables with more than 100 million rows, some with 500 million rows.
For the settings, currently only 1 table is running for full reload. while the table with the big data is running every few hours the below error appears.
As a result, the reload will restart back to 0 again :

detail log error:

Direct Path error

ORA-39767: finish is not allowed when unloaded stream data exists

Stream component 'st_1_ORA_TGT_MDWREPL_CUSTOMER_DEV_2' terminated

Stream component failed at subtask 1, component st_1_ORA_TGT_MDWREPL_CUSTOMER_DEV_2

Error executing data handler

Endpoint is disconnected

Direct Path error

ORA-03135: connection lost contact Process ID: 56526 Session ID: 646 Serial number: 65442

Thanks,

Trio

[SwathiPulagam · Accepted solution]

Hi @trioazhari ,

ORA-03135: connection lost contact

Cause: 1) Server unexpectedly terminated or was forced to terminate.
or 2) Server timed out the connection.

Action: 1) Check if the server session was terminated.
2) Check if the timeout parameters are set properly in
sqlnet.ora.

The solution should be implemented outside of replicate so please contact your DBA\Oracle support for more information related to the error. Please let me know if you have any questions.

NOTE: You can use parallel load to load a huge table: see UG get detailed usage (note not all source/target endpoints types support parallel load)

Below is the link for your reference:
https://help.qlik.com/en-US/replicate/November2021/Content/Global_Common/Content/SharedEMReplicate/C...

Thanks,

Swathi

[jimmyrekso]

Dear Swathi,

thanks a lot for your answer, would you like to tell me step by step configuration parallel load

thanks

Trio

[Steve_Nguyen]

@jimmyrekso 

you can refer to : 

 Parallel Load

https://help.qlik.com/en-US/replicate/November2021/Content/Global_Common/Content/SharedEMReplicate/C...

[trioazhari]

@Steve_Nguyen 

Thanks for the reference, let say we talking to example usage, here is configuration parallel load on my local system:

I'm curious about the meaning of segment boundaries and column paymentDate in define segment boundaries. The question is, is that true the boundaries '1' have a date range less than '2003-01-16', and the boundaries '2' have a date range less than '2004-01-16'?

Thanks,

Trio

[SwathiPulagam]

Hi @trioazhari,

Q: I'm curious about the meaning of segment boundaries and column paymentDate in define segment boundaries. The question is, is that true the boundaries '1' have a date range less than '2003-01-16', and the boundaries '2' have a date range less than '2004-01-16'?

A: Yes, you are correct.
 It is best to use the primary key column as it is unique and you can define equal segments based on PK min and max value.
If you don't have any unique columns with index then you can use date columns for the parallel load.
Now for your example make sure both segments will have an equal number of rows to get the best performance( means if 1st segment has 100 rows then 2nd segment also should have nearby 100 rows).

Thanks,

Swathi

[SwathiPulagam]

Hi @trioazhari,

Want to let you know if the target is Oracle, and the Direct Path option is selected then parallel load won't work it's because when you select direct path load into Oracle, there is no Parallelization supported. 

Thanks,

Swathi