| Same SCN in different Archive logs across nodes in a RAC environment. [message #157439] |
Thu, 02 February 2006 07:58  |
boomkap
Messages: 2
Registered: February 2006
|
Junior Member |
|
|
|
We were trying to perform an RMAN recovery in a multi node RAC environment. During recovery, we find that RMAN is trying to recover to the same SCN on archive logs generated by different nodes. Shouldnt SCN's be unqiue across nodes & thus be unique among archive logs generated by different nodes? What i am trying to understand is conceptually in a RAC environment why does this occur. Reading through the RAC concepts manual i dont see this being discussed.
|
|
|
|
| Re: Same SCN in different Archive logs across nodes in a RAC environment. [message #157484 is a reply to message #157439] |
Thu, 02 February 2006 13:39  |
boomkap
Messages: 2
Registered: February 2006
|
Junior Member |
|
|
Turns out that the default SCN generation scheme for RAC is Lamport. This generation scheme may not be fast enough for certain applications and can cause duplicate SCN's to be propogated in different archive logs by different threads. This can cause a read-inconsistency for a brief period od time. To avoid this one should change to the broadcast on commit SCN propogation. DBA's that administer a RAC environment need to be aware of the nature of the application and see if Lamport may be ok. For highly sensitive applications that need read consistency on all nodes as tx's are committed, broadcast on commit is the way to go.
Oracle has discussed the problem in this thread.
Duplicate SCN Identified In Different REDO LOG Threads
https://metalink.oracle.com/metalink/plsql/f?p=130:14:7230541823770998111::::p14_database_id,p14_docid,p14_show_header,p14_show_help, p14_black_frame,p14_font:NOT,260304.1,1,1,1,helvetica
I had this same issue posted on the Oracle forums. http://forums.oracle.com/forums/thread.jspa?threadID=359099&tstart=0
Thanks to Aron for finding the metalink URL.
|
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
