20.05.2008 04:41
isi
 
Oralce 9.2.0.8 + W2k3

Время от времени перестают исполняться job'ы (сложно сказать период но он измеряется не одним месяцем)
в Интете на эту тему нашел:

Bug: 3240280. JOBS SUDDENLY STOP ALTHOUGH NEXT_DATE <= SYSDATE

Кто нибудь имеет доступ на "металинк"- гляньте, может у кого и патч есть?

==========
просмотрел, видимо из той же темы https://olegon.ru/showthread.php?t=3764, но проблема не решена
20.05.2008 07:33
reddevil
 
PROBLEM: -------- DBMS jobs which have run successfully at their specified intervals in the past suddenly stop running since upgrade from 8174 to 9204. . SYSDATE passes the NEXT_DATE,NEXT_SEC of the job without the job running. The job has NO failures and is NOT broken. No errors from any Jnnn process or CJQ process. . . The job is still listed in DBA jobs but it never runs again (without manual intervention); . DIAGNOSTIC ANALYSIS: -------------------- Job will run for sometime if you 1) execute dbms_job.run(job#) manually. But eventually it will again, go into the state where it is never picked up again. . 2) alter system set job_queue_processes=0 and then set to > 1. Customer has job_queue_processes=30. But again, it will run for sometime, and then go into the state where it is never picked up again. . . Example query that customer uses to check jobs: . select schema_user,job,failures fails, what, broken, to_char(last_date,'dd-mon-yy hh24:mi') last_date, to_char(next_date,'dd-mon-yy hh24:mi') next_date, replace(replace(lower(interval),'trunc(sys'),',''ddd'')') interval from dba_jobs where next_date < = sysdate and broken = 'N'; SCHEMA_USER JOB FAILS ------------------------------ ---------- ---------- WHAT ------------------------------------------------------------------------------- - B LAST_DATE NEXT_DATE - --------------- --------------- INTERVAL ------------------------------------------------------------------------------- - DARTSYS 316842 0 response_time; N 30-oct-03 10:50 30-oct-03 11:05 sysdate+(1/96) . DND 317008 0 onguard_tools.load_from_feed_tables; N 30-oct-03 12:33 30-oct-03 12:38 sysdate + 5/1440 . As you can see, query only asks for those where next_date < = sysdate; however, these two jobs have not run, 0 failures, Not broken. . . It seems that the 2 jobs above are the 2 jobs that run at most frequent intervals. They have MANY jobs scheduled, and most run successfully, every now and then, one or 2 more jobs may enter this state, but then sometimes after manual intervention they continue to run. THe above 2 always seem to go into this state after a few successful executions. . . Had customer trace the CJQ process using oradebug with 10046, level 12. Trace is uploaded. Customer ran this for approx 20 minutes. And then during those same 20 minutes, customer queried dba_jobs_running and also queried dba_jobs to see what has a next_date < = sysdate. That log is also uploaded. (trace_log.txt) . WORKAROUND: ----------- 1) execute dbms_job.run(job#) manually. But eventually it will again, go into the state where it is never picked up again. . 2) alter system set job_queue_processes=0 and then set to > 1. Customer has job_queue_processes=30. . Workarounds not acceptable, it means to constantly have to watch the job queues and these are the jobs that run very frequently. Also, workaround is not permanent. Even with manual intervention, the jobs will eventually enter the state where next_date < = sysdate.
20.05.2008 08:17
isi
 
Получается что постоянного решения нет? Покопался ещё, выше 9.2.0.4 включительно баг вроде поправлен
Проверить не могу пока, джоб пересоздали, тот "пришили"...
Часовой пояс GMT +3, время: 21:31.

Форум на базе vBulletin®
Copyright © Jelsoft Enterprises Ltd.
В случае заимствования информации гипертекстовая индексируемая ссылка на Форум обязательна.