Skip to main content

Search

Welcome to Tibero GTS!

Issue with JOB Not Executing Automatically

Document Type | Troubleshooting

Category | Administration

Applicable Product Versions | 6FS06, 6FS07

Document Number | TADTS056

 

Issue

When executing a JOB manually, it runs normally, but there is an issue where it does not execute automatically.
 
 

Cause

Bug 1: Cause of continuous bitq occurrence

If a job scheduler thread has never recorded slog and a slog switch occurs, the existing implementation would skip the thread without special handling.
As a result, the agent considers that the log switch has not yet completed on that thread and continuously sends bitq signals to the job scheduler thread every fd check cycle (3 seconds).
 
 

Bug 2: Error in processing after receiving bitq

Although timeout recalculation was performed when awakened by receiving a signal, there was a logic error causing the timeout duration to gradually increase.
When awakened by receiving an unmasked bitq (e.g., log switch bitq), the recalculation logic itself was missing, causing the thread to go back to sleep with the initially set timeout value.
 
 

Solutions

Regarding the cause of bitq occurrence

Apply the patch for version 156652q or later to resolve the issue.
 

Improvement of bit processing logic

Apply the patch for version 225787 to resolve the issue.
The logic has been revised so that when awakened by receiving an unrelated bitq during a bitq read timeout wait, the timeout value is recalculated.
Caution
Apply the patch through technical support provided by Tmax Tibero.
Reference
As a temporary workaround, registering the following dummy job resolves the symptom.
DECLARE
job_no number;
BEGIN
DBMS_JOB.SUBMIT(job_no,
'dbms_lock.sleep(1);',
SYSDATE,
'SYSDATE+2/86400');
END;
/

Dummy job test
1. Check problem job (4)

SQL> SELECT * FROM DBA_JOBS where JOB='4';

 

      JOB SCHEMA_USER NEXT_DATE            INTERVAL             BROKEN   FAILURES WHAT                   INSTANCE LAST_DATE            THIS_DATE

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

        4          32 2024/04/06           TRUNC(SYSDATE+1)     N               0 KOSEAIS.SP_HRM_APPNT          0 2024/04/05

1 row selected.

 

2. Register and check 2-second job (22)

DECLARE

job_no number;

BEGIN

DBMS_JOB.SUBMIT(job_no,

'dbms_lock.sleep(1);',

SYSDATE,

'SYSDATE+2/86400');

END;

/

SQL> SELECT * FROM DBA_JOBS where JOB='22';

 

      JOB SCHEMA_USER NEXT_DATE            INTERVAL             BROKEN   FAILURES WHAT                   INSTANCE LAST_DATE            THIS_DATE

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

       22           0 2024/04/05           SYSDATE+2/86400      N               0 dbms_lock.sleep(1);           0 2024/04/05

 

3. Confirm automatic execution of problem job (4)


SQL> select * from dba_job_history where job = 4 order by end_date desc;


      JOB NAME       SCHEMA_USER   INSTANCE START_DATE           END_DATE             RESULT     COMMENTS

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

        4                     32          0 2024-04-05 18:02:27  2024-04-05 18:02:27  SUCCESS

        4                     32          0 2024-04-05 17:08:16  2024-04-05 17:08:16  SUCCESS

        4                     32          0 2024-04-05 16:55:02  2024-04-05 16:55:02  SUCCESS

        4                     32          0 2024-04-05 16:54:01  2024-04-05 16:54:01  SUCCESS

        4                     32          0 2024-04-05 16:53:00  2024-04-05 16:53:00  SUCCESS

        4                     32          0 2024-04-05 16:52:00  2024-04-05 16:52:00  SUCCESS
 
 
 
 

Related to