goldengate abended with no data found

先来看下报错ggserr.log：

2016-12-22 04:48:52  WARNING OGG-02544  Unhandled error (ORA-26787: The row with key ("SYNC_ID") = (bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3) does not exist in table M

SP.SYNC

ORA-01403: no data found) while processing the record at SEQNO 876, RBA 510082482 in Integrated mode. REPLICAT will retry in Direct mode.

2016-12-22 04:48:52  WARNING OGG-01154  SQL error 1403 mapping MSP.SYNC to MSP.SYNC OCI Error ORA-01403: no data found, SQL <UPDATE "MSP"."SYNC" x SET x."START_DAT

ESTAMP" = :a1,x."END_DATESTAMP" = :a2,x."STATUS_IND" = :a3,x."ERROR_MESSAGE" = :a4,x."LAST_SYNC_OFFSET" = :a5,x."LAST_SYNC_DATESTAMP" = :a6,x."USER_DEVICE_ID" = :a

7,x."CREATE_DATE" = :a8,x."UPDATE_DATE" = :a9,x."NIKE_PLUS_USER_ID" = :a10,x."UPM_USER_ID" = :a11,x."SYNC_RETRY_NUM" = :a12,x."APP_ID" = :a13 WHERE x."SYNC_ID" = :

b0>.

2016-12-22 04:48:52  WARNING OGG-01003  Repositioning to rba 510051955 in seqno 876.

2016-12-22 04:48:52  ERROR   OGG-01296  Error mapping from MSP.SYNC to MSP.SYNC.

Report at 2016-12-22 04:48:56 (activity since 2016-12-22 00:01:00)

From Table MSP.SYNC to MSP.SYNC:

#                   inserts:    203453

#                   updates:     73471

#                   deletes:         0

       #                  discards:        68 discard的记录有68条，是不是这68个记录都不符合要求还是第一个记录不符合要求，但这些是一个整体的事务，所以先去dirrpt里面看下。

From discard file:

OCI Error ORA-01403: no data found, SQL <UPDATE "MSP"."SYNC" x SET x."START_DATESTAMP" = :a1,x."END_DATESTAMP" = :a2,x."STATUS_IND" = :a3,x."ERROR_MESSAGE" = :a4,x

."LAST_SYNC_OFFSET" = :a5,x."LAST_SYNC_DATESTAMP" = :a6,x."USER_DEVICE_ID" = :a7,x."CREATE_DATE" = :a8,x."UPDATE_DATE" = :a9,x."NIKE_PLUS_USER_ID" = :a10,x."UPM_US

ER_ID" = :a11,x."SYNC_RETRY_NUM" = :a12,x."APP_ID" = :a13 WHERE x."SYNC_ID" = :b0>

Aborting transaction on /gg/app/ggadmin/product/12.1.2.1.0/dirdat/replicate/pdsp/r1 beginning at seqno 876 rba 510051955

error at seqno 876 rba 510082482

Problem replicating MSP.SYNC to MSP.SYNC

Record not found

Mapping problem with compressed update record (target format)...

*

SYNC_ID = bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3

…………

Continuing to discard records up to the last discarded record from action ABEND

Operation discarded from seqno 876 rba 510051955

Aborted compressed update from MSP.SYNC to MSP.SYNC (target format)...

*

SYNC_ID = f00934c3-3c04-4928-ae90-b2a01e0c7572

Operation discarded from seqno 876 rba 510052452

Aborted compressed update from MSP.SYNC to MSP.SYNC (target format)...

*

SYNC_ID = 206fb814-cd8d-4add-9b06-d8e4fecf8976

选了discard file前面部分，后面还有很多record就不一一看了。

先来看报错，说是记录不存在。

于是先在BI目标库看看，

SQL> select sync_id from msp.sync where sync_id in ('bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3','206fb814-cd8d-4add-9b06-d8e4fecf8976','f00934c3-3c04-4928-ae90-b2a01e0c7572');

SYNC_ID

----------------------------------------

206fb814-cd8d-4add-9b06-d8e4fecf8976

f00934c3-3c04-4928-ae90-b2a01e0c7572

可见第一条记录的确不存在，那源库存在吗？

源库PDSP：

SQL> select sync_id from msp.sync where

2  sync_id in ('bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3','f00934c3-3c04-4928-ae90-b2a01e0c7572','206fb814-cd8d-4add-9b06-d8e4fecf8976','3346821d-40e2-4ab8-a650-a68ffd66e9b2');

SYNC_ID

----------------------------------------

206fb814-cd8d-4add-9b06-d8e4fecf8976

3346821d-40e2-4ab8-a650-a68ffd66e9b2

bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3

都存在

那么就得insert这条不存在的记录到BI里面取，然后起R1_SP，看看还有造成abended的记录吗（因为这一个事务包含着68条记录，为什么能确认这是一个事务，可以logdump进去根据事务标识符判断，这里就略过了）

Bi遗失的记录在源库查询：

SQL> select * from msp.sync where sync_id='bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3';

SYNC_ID                                  START_DAT END_DATES STATUS_IND

---------------------------------------- --------- --------- ----------

ERROR_MESSAGE

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------LAST_SYNC_OFFSET LAST_SYNC USER_DEVICE_ID                           CREATE_DA UPDATE_DA NIKE_PLUS_USER_ID                        UPM_USER_ID                        SYNC_RETRY_NUM     APP_ID

---------------- --------- ---------------------------------------- --------- --------- ---------------------------------------- ---------------------------------------- -------------- ----------

bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3     11-OCT-16                    0

proc_fuelband_wrapperclass sets status_ind to 0.

0 01-JAN-70 e3659704-686e-4e25-9185-41e67aef57eb     11-OCT-16 22-DEC-16 b033e11a-a94a-4fc7-b153-c12f720a8278     9745824936                              0 335

可以在toad上查找这条记录，然后导出insert格式的语句，如下：

 

会生成如下：

SET DEFINE OFF;

Insert into MSP.SYNC

(SYNC_ID, START_DATESTAMP, END_DATESTAMP, STATUS_IND, ERROR_MESSAGE,

LAST_SYNC_OFFSET, LAST_SYNC_DATESTAMP, USER_DEVICE_ID, CREATE_DATE, UPDATE_DATE,

NIKE_PLUS_USER_ID, UPM_USER_ID, SYNC_RETRY_NUM, APP_ID)

Values

('bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3', '11-OCT-16', '', '0', 'proc_fuelband_wrapperclass sets status_ind to 0.',

'0', '01-JAN-70', 'e3659704-686e-4e25-9185-41e67aef57eb', '11-OCT-16', '22-DEC-16',

'b033e11a-a94a-4fc7-b153-c12f720a8278', '9745824936', '0', '335');

COMMIT;

然后BI库跑一下，然后起R1_SP即可。

另一个操作版本（不建议）：

没插数据，直接加HC参数，直接起，但是BI库却始终看不到这条记录。因为HC已经把这条记录给skip了。

但是我记得HC参数对于update会转换成insert语句，却没看到这条记录也就说明没转换。

所以还得去看看官方文档对于HC在各个场景下的策略。

如下就是等到R1_SP都AT EOF了，都没看到这条记录。

send r1_sp status

Sending STATUS request to REPLICAT R1_SP ...

Current status: At EOF

Sequence #: 876

RBA: 518830790

0 records in current transaction

SQL> select * from msp.sync where sync_id='bd7a9e64-0301-4cbf-bc10-9a8c0a231fc3';

no rows selected

SQL> /

no rows selected

简单的案例想说明的就是：

In world of Golden Gate you can do anything only make sure that you’re not losing any transaction or record.