gerd_renz3
VIP Member
Hi List,
for long it has been a problem that UBEs do not treat properly (database)
runtime errors.
Errors may show up in logs during execution, however, many times the UBE
still finishes with status 'D' and the logs get automatically deleted.
(Leaving debugging on is not an option for a production shop.)
To avoid inconsistency in our Database we adopted some measures to catch
errors when they occur. During execution we check for errors (using the
Unix 'grep' command) within the logs. If an error shows up we copy the log
in a separate directory for further examination and to save it from
deletion.
We are locking for strings like: missing, Failed, ORA-, JDB, etc.
This seems a quite primitive method to us but we have not found anything
better. Actually we have been able to reduce the occurance of these errors
to almost zero. Now we are implementing new suites and we expect, at least
in the beginning, some errors to show up again.
Our UBEs are being executed at night by operators. A DBA is not always
present.
We are in production on B733.1 and about to go live on Xe. Has anything
changed in Xe in this respect? Does OW treat database errors in a more
efficient way now?
Does anybody else worry about this at all? How are you treating it?
Thanks for your answers, Gerd
B733.1 -> Xe, AIX 4.3.3, Oracle 8.1.7
for long it has been a problem that UBEs do not treat properly (database)
runtime errors.
Errors may show up in logs during execution, however, many times the UBE
still finishes with status 'D' and the logs get automatically deleted.
(Leaving debugging on is not an option for a production shop.)
To avoid inconsistency in our Database we adopted some measures to catch
errors when they occur. During execution we check for errors (using the
Unix 'grep' command) within the logs. If an error shows up we copy the log
in a separate directory for further examination and to save it from
deletion.
We are locking for strings like: missing, Failed, ORA-, JDB, etc.
This seems a quite primitive method to us but we have not found anything
better. Actually we have been able to reduce the occurance of these errors
to almost zero. Now we are implementing new suites and we expect, at least
in the beginning, some errors to show up again.
Our UBEs are being executed at night by operators. A DBA is not always
present.
We are in production on B733.1 and about to go live on Xe. Has anything
changed in Xe in this respect? Does OW treat database errors in a more
efficient way now?
Does anybody else worry about this at all? How are you treating it?
Thanks for your answers, Gerd
B733.1 -> Xe, AIX 4.3.3, Oracle 8.1.7
