Redolog doubt in RAC

Question

Hi,
I have small doubt regarding redolog corruption/deletion in RAC Instance:
OS: RHEL 4.5
Oracle: 10.2.0.1
DB_NAME=RAC ( 2 Instances, RAC1 and RAC2 are pointed to this DB).
One user (User1) is connected to DB using RAC1 Instance having redolog groups 2 (having each one member: redo1a.log and redo2a.log). I did logswitch freshly and Presently current redolog is redo2a.log. User1 created emp table and inserted 20 records and commited. note that, still current redolog is redo2a.log and there is no log switch. There are no users connected to DB other than user1.
Now, I shutdown abort the RAC1 instance and User1 session is switched to RAC2 Instance. When he queries the select * from emp table, he is getting 20 records? how come he gets? the inserted records are still in redolog2a.log of RAC1 instance and commit statement is also still there in redolog and thre is no log switch, instance RAC1 is shutdown abruptly. How come he gets 20 records, when he connected to RAC2 instance?
Please explain me, how it is possible? please execuse me, if there is anything wrong with my question.

Answer 1

Whether it is RAC or non RAC system DBWn process writes the dirty buffers to disk.
The DBWn process writes dirty buffer to disk under the following conditions:
1. When a checkpoint is issued.
2. When a server process cannot find a clean reusable buffer after scanning a threshold number of buffers.
3. Every 3 seconds
if still not written to disk when you start database next time media recovery will be completed.
Where in RAC environment it works on concept of GLOBAL CACHE. Whichever changes happening to 1st node same will be update to 2nd node as well. when node 1 goes down abruptly 2nd node identifies it and does the media recovery. so even when user again connects to 2nd node still data will be available.
Anil Malkai