用户session会不会因为RAC节点down掉而自动转移?

duxzh

最初由 jiangjs888 发布
[B]假设有两个RAC实例,主机Server1,Server2,实例名ora9i

ORA9I =
  (DESCRIPTION =
    (FAILOVER=on)
    (LOAD_BALANCE=on)
    (ADDRESS_LIST =
      (ADDRESS = (PROTOCOL = TCP)(HOST = Server1)(PORT = 1521))
      (ADDRESS = (PROTOCOL = TCP)(HOST = Server2)(PORT = 1521))
    )
    (CONNECT_DATA =
      (SERVICE_NAME = ora9i)
      (FAILOVER_MODE=(TYPE=SESSION))
    )
  )

注:TYPE=SESSION表示会话级FAILOVER,=SELECT表示select语句failover [/B]

强!
那天做个试验.

xzh2000

Services Entries  When you configure high availability services using the DBCA Services page, then the DBCA creates net service entries similar to the following. The three services in the following examples, db_svc1, db_svc2, and db_svc3, have TAF policies of NONE, BASIC and PRECONNECT respectively.
db_svc1.us.acme.com=
  (description =
    (address=(protocol=tcp)(host=node1-vip)(port=1521))
    (address=(protocol=tcp)(host=node2-vip)(port=1521))
    (load_balance=yes)
    (connect_data=
        (server = dedicated)
        (service_name = db_svc1.us.acme.com)
  )
  )

db_svc2.us.acme.com=
  (description=
    (address=(protocol=tcp)(host=node1-vip)(port=1521))
    (address=(protocol=tcp)(host=node2-vip)(port=1521))
    (load_balance=yes)
    (connect_data =
        (server = dedicated)
        (service_name=db_svc2.us.acme.com)
        (failover_mode =
        (type=select)
        (method=basic)
           (retries=180)
           (delay=5)
     )
     )
     )

db_svc3.us.acme.com=
    (description=
      (address=(protocol=tcp)(host=node1-vip)(port=1521))
      (address=(protocol=tcp)(host=node2-vip)(port=1521))
      (load_balance=yes)
      (connect_data=
        (server=dedicated)
        (service_name=db_svc3.us.acme.com)
        (failover_mode=
        (backup=db_svc3_preconnect.us.acme.com)
        (type=select)
        (method=preconnect)
        (retries=180)
        (delay=5)
   )
   )
   )


Services Entries (continued)  When a service has a TAF policy of PRECONNECT, then a service_name_preconnect net service entry is also created as in the following example:
db_svc3_preconnect.us.acme.com =
  (description =
    (address = (protocol = tcp)(host = node1-vip)(port = 1521))
    (address = (protocol = tcp)(host = node2-vip)(port = 1521))
    (load_balance = yes)
    (connect_data =
      (server = dedicated)
      (service_name = db_svc3_preconnect.us.amce.com)
      (failover_mode =
        (backup = db_svc3.us.acme.com)
        (type = select)
        (method = basic)
        (retries = 180)
        (delay = 5)
      )
    )
  )

External procedures  An entry for connections to external procedures. This enables an Oracle Database 10g database to connect to external procedures.
extproc_connection_data.us.acme.com=
(description=
  (address_list=
    (address=(protocol=ipc)(key=extproc0))
  (connect_data=
    (sid=plsextproc)))

duxzh

更强!谢谢大家了!!

西门吹牛

最初由 duxzh 发布
[B]谢谢piner!
那么rac的最大优势在于性能的提高吧?
[/B]

不对。rac的优势在于高可靠性。

lc7888

我的理解是当应用程序的一个session链接到A 节点时，如果该节点此时突然宕机，这个session会被中断，但是应用程序会自动去链接另外一个节点，也就是用户需要重做这个session的操作……
不知这样理解是否正确？

husthxd

RAC环境下Oracle Net Configuration for Failover说明
来源于下面的ebook
McGraw.Hill.Oracle.Database.10g.High.Availability.with.RAC.Flashback.and.Data.Guard.Apr.2004.eBook-DDU

Connect-Time Failover
We begin with connect-time failover, which is defined as a failed initial connection attempt that must be retried against a different address. Consider the following entry in a client's tnsnames.ora:
代码:



GRID =

  (DESCRIPTION =

    (ADDRESS = (PROTOCOL = TCP)(HOST = rmscvip1.us.oracle.com)(PORT = 1521))

    (ADDRESS = (PROTOCOL = TCP)(HOST = rmscvip2.us.oracle.com)(PORT = 1521))

    (CONNECT_DATA =

      (SERVICE_NAME = grid)

    )

  )


Given the above entry for the alias called GRID, when a client attempts to connect, the first address in the list will be the one initially tried-in this case, rmscvip1. Should the initial connection attempt to rmscvip1 be unsuccessful, the next address in the list will be tried. This is known as a connect-time failover. You can add as many entries to the list as you have available nodes to accept connections. Note also that the entries in the list do not have to be RAC nodes. The above may apply in a Streams or advanced replication environment, or one of the entries in the list may be a standby database-either physical or logical, as long as there is an available service defined and running on that node. The other criteria, of course, is that the addresses in the list will all allow the client to get to the data that is needed by the application. In the case of a RAC environment, we know that the data is always the same, as it is the same database. In the case of Streams, advanced replication, and physical or logical standby, whether or not the data is accessible depends on how these options are configured.

Transparent Application Failover
Transparent Application Failover, or TAF, refers to the failover of a connection that has already been made when a node or instance goes down. This is controlled by the parameter FAILOVER=ON, in the tnsnames.ora. By default, this parameter will be set to on, if not defined. Setting it to OFF will disable failover. When it is enabled, there are two types of failover, defined as part of the FAILOVER_MODE parameter: session (the default), or select. With TYPE=SESSION, should an instance fail after a session is connected, that session will be failed over to another available address in the list, without the user needing to reconnect. However, any SQL statements that were in progress will need to be reissued. With TYPE=SELECT, if the session fails over in the middle of a query, the query will be restarted after the failover. Consider the following modification to the CONNECT_DATA section of the tnsnames.ora file:
代码:



(CONNECT_DATA =

      (SERVICE_NAME = grid)

      (FAILOVER_MODE =

        (TYPE = SELECT)

        (METHOD = BASIC)

      )

  )


In this case, should a failure occur, the failed over session will keep track of the number of rows retrieved so far, so that when the query is restarted, the session will discard those rows that have already been returned.

husthxd

另外的参数说明:

1.METHOD
Determines how fast failover occurs from the primary node to
the backup node:
代码:



basic: Set to establish connections at failover time. This

option requires almost no work on the backup server until

failover time.

preconnect: Set to pre-established connections. This

provides faster failover but requires that the backup

instance be able to support all connections from every

supported instance.



2.TYPE
Specify the type of failover. Three types of Oracle Net failover
functionality are available by default to Oracle Call Interface
(OCI) applications:
代码:



session: Set to failover the session. If a user's connection is

lost, a new session is automatically created for the user on

the backup. This type of failover does not attempt to recover

selects.

select: Set to enable users with open cursors to continue

fetching on them after failure. However, this mode involves

overhead on the client side in normal select operations.

none: This is the default. No failover functionality is used.

This can also be explicitly specified to prevent failover from

alonemo

So Stronger

duxzh

继续感谢大家!
希望还有高人共享自己的知识!