22101 Segmentation Fault - core dumped & Parallel job re

[apkselvam]

Hi All,

We are facing performence problem in a job which will execute merge query in Oracle database. I tried to modify the query to split the volume & run it paralley.Below is the original query,

MERGE INTO #$DM_SCHEMA_PREFIX#PART_FACT a
USING #$DM_TMP_SCHEMA_PREFIX#TMP_PART_FACT b
ON ( a.dealer_key = b.dealer_key
AND a.part_key = b.part_key
AND a.time_dim_key = b.time_dim_key
AND a.purchase_order_number = b.purchase_order_number
AND a.sales_ord_inv_flag = b.sales_ord_inv_flag
AND a.invoice_line_number = '-1001'
AND a.purchase_order_number = '-1001'
AND a.order_line = '-1001')
WHEN MATCHED THEN
UPDATE
SET a.sales_quantity = b.sales_quantity,
a.lost_sales_quantity = b.lost_sales_quantity,
a.source_system_id = b.source_system_id,
a.update_user_id = b.update_user_id,
a.update_date = b.update_date, a.tstamp = b.tstamp
WHEN NOT MATCHED THEN
INSERT (a.dealer_key, a.part_key, a.time_dim_key, a.sales_quantity,
a.purchase_order_number, a.lost_sales_quantity, a.create_date,
a.source_system_id, a.create_user_id, a.tstamp, a.sales_ord_inv_flag)
VALUES (b.dealer_key, b.part_key, b.time_dim_key, b.sales_quantity,
b.purchase_order_number, b.lost_sales_quantity, b.create_date,
b.source_system_id, b.create_user_id, b.tstamp, b.sales_ord_inv_flag)

Modified query:
*************

MERGE INTO #$DM_SCHEMA_PREFIX#PART_FACT a
USING ( SELECT * FROM #$DM_TMP_SCHEMA_PREFIX#TMP_PART_FACT WHERE MOD(DBMS_ROWID.ROWID_BLOCK_NUMBER(ROWID)-2,4)=0 ) b
ON ( a.dealer_key = b.dealer_key
AND a.part_key = b.part_key
AND a.time_dim_key = b.time_dim_key
AND a.purchase_order_number = b.purchase_order_number
AND a.sales_ord_inv_flag = b.sales_ord_inv_flag
AND a.invoice_line_number = '-1001'
AND a.purchase_order_number = '-1001'
AND a.order_line = '-1001')
WHEN MATCHED THEN
UPDATE
SET a.sales_quantity = b.sales_quantity,
a.lost_sales_quantity = b.lost_sales_quantity,
a.source_system_id = b.source_system_id,
a.update_user_id = b.update_user_id,
a.update_date = b.update_date, a.tstamp = b.tstamp
WHEN NOT MATCHED THEN
INSERT (a.dealer_key, a.part_key, a.time_dim_key, a.sales_quantity,
a.purchase_order_number, a.lost_sales_quantity, a.create_date,
a.source_system_id, a.create_user_id, a.tstamp, a.sales_ord_inv_flag)
VALUES (b.dealer_key, b.part_key, b.time_dim_key, b.sales_quantity,
b.purchase_order_number, b.lost_sales_quantity, b.create_date,
b.source_system_id, b.create_user_id, b.tstamp, b.sales_ord_inv_flag)

Difference between the above queries is :,

We included the below statement , but the job is aborting saying
Parallel job reports failure (code 139).

SELECT * FROM #$DM_TMP_SCHEMA_PREFIX#TMP_PART_FACT WHERE MOD(DBMS_ROWID.ROWID_BLOCK_NUMBER(ROWID)-2,4)=0 )

Is there any environment variable do I need to set?

Thanks in advance.

[ArndW]

Is there any more to the error message, the Segmentation fault and abort message don't narrow down the issue at all. Also, does the new SELECT function correctly from your preferred tool?

[apkselvam]

[quote="ArndW"]Is there any more to the error message, the Segmentation fault and abort message don't narrow down the issue at all. Also, does the new SELECT function correctly from your preferred tool? ...[/quote]

Thanks for response Arndw, Same query is working fine in Toad, but I tried to run through it say parallel job aborts code 139.

[ArndW]

The 139 doesn't help much. Somewhere in the error message there will be a SQL code or other more specific message.

[miwinter]

OP - You are on v7x, right? I've not seen an error like this before that's all but I know there was a newly produced error code reference of some ilk for v8.

Furthermore, when you say

the job is aborting saying Parallel job reports failure (code 139)

What is "saying" this? Is that a line out of the job log or some calling script or otherwise?

[sjfearnside]

I experienced the 139 error and it was a problem with my odbc.ini file.
err msg => Parallel job reports failure (code 139)

[apkselvam]

[quote="miwinter"]OP - You are on v[u]7[/u]x, right? I've not seen an error like this before that's all but I know there was a newly produced error code reference of some ilk for v8.

Furthermore, when you say

[quote]the job is aborting saying Parallel job reports failure (code 139)[/quote]

[u]What[/u] is "saying" this? Is that a line out of the job log or some calling script or otherwise?[/quote]
Yes you are right am using V7.5.1 , it error code from job log. I tried using ODBC stage still the aborts. I think there would be a problem with Datatstage version 7x series, When a trigger goes from datastage to database it aborts> If I run same query in sqlplus it goes fine.

[hema177]

Hi,

I am facing the same issue when placing an update call to Oracle via Oracle Enterprise stage

Contents of phantom output file =>
509 Segmentation Fault - core dumped

The query works fine on toad.

If you know a workaround to this please let me know.

Thanks,

Hema

[Sainath.Srinivasan]

509 is different from 139.

What is the SQL ?

Did you include any parameters ?

Did you supply all values ?

[ray.wurlod]

I am facing the same issue ...

No you're not. Please start a new thread.

[hema177]

How would you know I am not?

Anyway, the resolution seems to be related to the length of the sql being passed via the Oracle Enterprise Stage. I shortened the sql and the job now works.

[ray.wurlod]

How would you know I am not?

Because you are getting a different error code.