|
|
My recent real story about database crashs/disaster and recoveries. Just for reference to protect your data safely and save your, your company and our China's money.
I am a DBA responsible for Oracle and Sybase databases. There are about 30 servers. All of the servers(machines) are protected and supported by UPS system.
one day morning, there were two times of power failures and our UPS did not switch on. i.e. no battery backups for these servers did not work at all.
After power recoverd, 1 oracle database (8.1.7.2, producation) complained "file 1 need media recovery.." (you know it is system.dbf file)
4 sybase databases(producation) marked suspect i.e. need recovery.
As to Oracle database, we run it on archived log mode and have daily backup. Usually keep 5-day backup online.
At begining, expect the online logs are intact, I tried to recover it using 'alter database recover datafile...' to recover each dirty datafiles. But I failed , Oracle still complained something.
Finally as last resort, using our backups I recovered it point in time. The whole recovery procedure just took 5 hours. I was happy, my boss was happy and my users were happy too.
Just for your reference,our backup methods(for production).
0. have an exclusive backup server
1. Oracle keeps its archived log; Sybase dump trans
2. fully backup or increment backup based on your size of databases.
3. hourly backup the archive log and transaction log
4. keep 5-day backups online i.e.on disk
5. move backups to tape
6. 3-days bring tapes to off-site location i.e.other places
So as to be a DBA, no matter from your job security point of view or from your responsibility( or moral) point of view, you MUST have complete and tested methods to make sure your databases safe or recoverable in any circumstances
Personaly, very thanks alantany to release his disaster to all of us. We have a chance to learn this kind of thing that are rarely happened and also sincerely hope alantany can work more to make your data more security . |
|