I can honestly say that the past week has just been horrible for me, work-wise. I'm still having some growing pains with my roll-out of Oracle RMAN for my company's backup/recovery strategy. As anyone knows, you are only as good as your last backup. And I found out that some of my early backups were just not good.
First I was trying to do an RMAN duplication to a development instance which bombed due to three mysteriously missing archive logs. I spent two days trying to understand a) where in the hell those logs were, and b) why RMAN even needed them. It turns out RMAN needs them because they were created during the incremental backup on that particular night. Otherwise it would only have needed the incremental backupset itself, no logs. I never did get those logs, instead asking the developers to pick another date.
That second duplication failed at the very end with date format string problems. I spent 3 freaking days trying to understand it. I'm sure you can find my desperate pleas for answers on the internet. Normally, date format errors involve the user specifying an UNTIL TIME parameter in an improper format, and RMAN will complain right away. This problem wouldn't complain until after restore and recovery were complete. In my case, a full restore and recovery can take over 10 hours.
Well it turns out that the problem as a mismatch in the NLS_DATE_FORMAT from the target instance ('DD-MON-RR') to the auxiliary instance ('YYYY-MM-DD HH24:MI:SS'). I've since unset NLS_DATE_FORMAT on the auxiliary instance, and sworn to just use the "TO_DATE()" function instead.
And while I was at my most frustrated and lowest point, my 1-year-old son dumped a full mug of coffee onto my laptop. Luckily my hard-drive was salvaged and transplanted.
In the end everything is working (except my old laptop), and I know why it wasn't before. Thankfully it was "only" development restores that were held up.
Plus I've turned into a complete social networking loser, with accounts now on MySpace, Facebook, and now LinkedIn.