ORA-01033 Error cannot be traced in the alert log or v$views

Question

Hello There,
I'm hoping you can shed some light on what seems a rather odd occurrence on our Production Oracle Instance.
Before i elaborate on the nature of the problem, I must confess that i am not an Oracle DBA and have been compelled to post this query since i have no joy from the in house DBA community on the origins of this error.
We've had an ORA-01033 error being issued by our ETL installation (deployed on a Linux machine) a couple of days ago whence trying to extract some data from the Oracle instance (during overnight DataWarehouse loads) which has consequently aborted the loads necessitating a cleanup.
This event has already occurred twice before, the first time being a month ago when the event was captured in the v$instance (using the Startup time column where the timestamp corroborates to the first time we experienced this issue and also ackowledged by the DBA team) and also in the V$Dataguard_status.
Since then, This error has been generated twice although there seems to be no evidence of this in either the Oracle alert log(as confirmed by DBA team) or in any of the v$views (as pointed out by Tom in an earlier post) such as DBA_LOGSTDBY_HISTORY, v$logstdby_state, $logstdby_stats, dba_logstdby_log, dba_logstdby_events,v$dataguard_status, v$dataguard_stats, v$dataguard_config,v$database,v$database_incarnation, v$managed_standby, v$standby_log, v$instance. I searched these views since i suspected a latency issue during a failover which could be the reason for the
ORA-01033 but found nothing.
The DBA team have pretty much disowned this issue since they claim to not have any actual evidence of this from the logs and this is the crux of the matter here.
The Problem i have as the downstream "recepient" of this error is to prove to the DBA team that this is indeed a genuine issue with the Oracle instance affecting it's availability and concomittantly affecting the DW loads.
FYI, The Oracle instance is in Failover mode, so it's swiftly back online after bombing out for a few seconds.
Also, I don't have access to the Alert log as it's a Production environment and employs restricted access policy to the server.
Having said that, Is there anything else besides the obvious ORA errors that should be looked for in the Alert log?
Where else can ORA-01033 errors be traced/traced in the Dictionary besides the Alert Log?
Thoughts??
Regards

Answer 1

Thank You John for that query, but i'm on V10.2 and this view is relevant for 11g and beyond i believe.
Perhaps there is an equivalent for V10g?
I am also a bit bemused by the comment earlier about no trace being left behind if the DBA's performed a manual restart, surely it doesn't matter how the DB is restarted, the event is captured in the dictionary?
In the meantime, I've got a copy of the Alert log and have found redo log issues (DBWR/LGWR) very close to the time (a min after ORA-01033) of the shutdown event.
ALTER SYSTEM ARCHIVE LOG
Thread 1 cannot allocate new log, sequence 117732
Checkpoint not complete
I've looked into this a fair bit and this error apparently causes Oracle to suspend all processing on the database until the log switch is made.
Is this the causal link for my issue?
Does " suspend all processing on the database " actually translate into an actual ORA-01033 error (or some form of) when the ETL application is trying to connect to and extract data from the Oracle database at that time?
Edited by: shareeman on 16-Oct-2012 03:50