Re: Oracle8.1.7 snapshot refresh failed without any error message

jshen.cad_at_gmail.com wrote:
>>> There had been refresh failures before. At those failure time, I
>>> cound determine refresh failure by " select job, failures, last_sec,
>>> next_sec from user_jobs".
>> Hmmm - does not sound like a stable system. Ever found
>> out what causes these errors?
>
>
> No. Usually it just fails, we do not keep the alert.log . To my
> experience, there has been 4 times of refresh failure this year.
>

And why do you think errors on jobs will end up in the alert log file?
They should be logged by the pl/sql block that is executed when the MV refreshes.

Manually executing the procedure, that is called by the job will reveal the error.

>
> In fact, there is NO error on refresh failure in alert.log . Before
> failure happened,
> there has been several Archive failure log in alert.log, but there
> exist successful archive after failure log.
>
> Those Error looks like :
>
> Sun Dec 2 03:10:27 2007
> ARC0: Beginning to archive log# 13 seq# 361643
> ARC0: Failed to archive log# 13 seq# 361643
> ARC0: Beginning to archive log# 15 seq# 361644
> ARC0: Failed to archive log# 15 seq# 361644
>

Dangerous - should be an entry in a trace file, explaining why the archiver failed - disk full?

>
>> Suggest you check your time settings on the servers (best to let
>> then synchronize among each others, and have one synchronize, using
>> the nntp (time) protocol.

> Our server time are synced to a NTP server.

OK.

-- 
Regards,
Frank van Bortel

Top-posting is one way to shut me up...