Event log shows " Log parser error 5" (4217943)

"show capture detail" shows "Statistic data not available" - Found on RAC system. The seqno entry for the actid xx(the current actid) for one of the node is different. Example 114141 whereas in Capture log it is somewhere in the of 112868.

Capture log:

ocap 2011-01-04 00:43:51.410788 29641 3755654032 Log parser error 5

ocap 2011-01-08 04:41:46.151372 29641 3755654032 Block zero has seqno 112868, should be 112860

For some unknown reason, capture was getting lots of incomplete/transient block when it read online redolog (even on block 0 and block 1), which caused capture to generate lots of Log Parser error message and sometimes to core.

Fix ST81900 attempts to address the partial/transient block problem by adding the checksum-based redolog ensure to make sure that the block read is complete. Also, it's observed that sometimes capture was stuck inside the sleep (nanosleep()) call and never returned. To address the problem with getting struck in sleep call, capture will use select call instead.

Resolution 1:

Upgrade to 7.5.2.10 which has fix or latest 7.5.2.x patch

Resolution 2:

Request oneoff patch ST81900

Workaround:

1. Resync the tables
2. Update the source table Shareplex actid to start capture processing from the archive desired or available:
3. Stop capture
4. Confirm that capture is stopped issuing sp_ctrl> show
5. Note the actid of the active confif file by issuing sp_ctrl> list config (
6. Login to Oracle on the source db as shareplex user
7. Update the <splex user>.shareplex_actid

update shareplex_actid set SEQNO=<sequence# to start capturing from that is available), LOG_START_OFFSET=0, OFFSET=0 where ACTID= <current activation id >;

commit;

NOTE: If RAC system , update all SEQNO of the instance nodes to the available archived log and LOG_START_OFFSET=0, OFFSET=0 where ACTID= <current activation id >

Commit;

1. sp_ctrl> start capture