Issuing blanket statements that imply dire problems without any specifics is the lesson we should have all learned from Chicken Little.
In what cases can it corrupt data? Versions? O/S? Single Instance or RAC? Specific scenarios? At the very least reference a metalink note or bug #. I have no doubt that data can get corrupted, perhaps even by a shutdown abort, most likely caused by a bug in the Oracle software. Shutdown abort has been used for years, quite successfully as many of us on this list can attest. I have no qualms about using it because I understand what it does and does not do.
And...is not the point of shutting down to put the instance in an unusable state (not being up is certainly not usable). The point of shutdown normal/transactional/immediate is to put the instance into an unusable state as well.
I have also noticed two other disconcerting items in the paper. There is a reference to a server manager session (about 1/2 way down on the last page) and there is no reference to any date nor oracle release. It makes me wonder how old this paper is. I do recall that in 7.0 days, shutdown abort was frowned upon...though no one could ever explain why. But as the most junior of DBAs, I took the word of my more learned colleagues.
Regards, Daniel Fink -- Daniel Fink Oracle Performance, Diagnosis and Training OptimalDBA http://www.optimaldba.com Oracle Blog http://optimaldba.blogspot.com Joel.Patterson@xxxxxxxxxxx wrote:
I was reading (and contributed my .02£), to the forum discussion awhile ago concerning shutdown abort yet came across this advice and warnings of possible corruption still being advertised. Does this lend credit to the institution, or take points away? The leanings where very much favorable to abort with much in depth analysis. http://www.embarcadero.co.uk/resources/tech_papers/oracleorphan.pdf If so, one wonders.... (not that I have any intention of utilizing there services in the first place... it is simply academic). -- http://www.freelists.org/webpage/oracle-l