Re: possible nomination for oracle wtf
Date: Fri, 24 Jul 2009 10:26:23 +0200
John Hurley schreef:
> On Jul 23, 1:05 pm, gazzag <gar..._at_jamms.org> wrote:
>> I think you completely missed the point, John. As others have >> commented, she was pointing out the value of proper error reporting: >> >> "....the basics of error handling should not be ignored – When an >> error occurs, we must know what was the error." >> >> It reminds me of the circumstances in which I entered the world of the >> Oracle DBA. Many years ago, Oracle 7.3.4, in fact, I was a contractor >> in a government department. At the time, the incumbent Oracle DBA was >> away for a week and we had a problem that could possibly have required >> us to revert to the previous night's cold backup. I was tasked with >> locating the relevant tapes and checking their contents. I tried the >> tape from the previous night's backup and my heart sank as it appeared >> to be empty. The tape from the night before was empty aswell. As I >> worked backwards through tape after tape, it became apparent that we >> had no backup whatsoever. >> >> I located the script that was run by cron every night and examined >> it. There, the very last line in the script simply was: >> >> echo "Backup completed." | mail <list_of_recipients> >> >> There was no logic within the script - that line would be executed >> regardless. Fortunately we found a workaround to the issue, so no >> real harm was done. Apart to the DBA's career, naturally. When the >> then-DBA returned from holiday he was "promoted sideways" and I was >> treated to approximately six weeks worth of training with Oracle :-)
> The part especially for Oracle WTF was the code in the WHEN OTHERS
> logic ... basically job failed not catching how/why and no RAISE
> But it is at least amusing to me the politics involved in having
> Oracle ACE's employed ( that might or might not have a clue ... it
> probably varies a lot person by person ) in organizations that allow
> stuff like this to either almost run in production or really get
> installed into production.
> It could be the factor "well we know better but we do not have any
> management willing to stop stuff like this from happening".
> Also amusing is the idea that this project is so big and so important
> and ... well let's allow the DBA involved to take a 3 week vacation as
> it goes live ... that sounds like a recipe for fireworks if any
> problems develop.
It's not uncommon to see people go on vacation when a product goes live. Most times they planned to go on vacation AFTER the product goes live, but since all projects have delays, planned vacations always are at the wrong moments.... I've seen that happening numerous times....
Shakespeare Received on Fri Jul 24 2009 - 03:26:23 CDT