Failover of applications TimesTen

Similar Questions

• Questions about Failver customer and rapid failover

  have some questions about customer and quick Faiolver of Oracle Database HA failover. Before asking these questions, I want to explain my environment. Here are the details.
  
  
  
  -We have two physical locations called "ABC" and "PQR".
  -ABC is the main site.
  -PQR is the backup site.
  -In ABC, we have the database to Oracle RAC (11.2.0.2) with two nodes.
  -In the PQR, we have only one stand-alone server (11.2.0.2) database with ASM. This isn't a RAC.
  -Data Guard has been configured between ABC and PQR and it works as expected.
  -Please note that we have a license for Active Data Guard.
  -We have products of Oracle Identity Management to ABC and PQR and they will use the RAC database as a primary database that is in the CBA.
  -We have not yet set up a Data Guard Broker.
  
  We want to achieve under objectives:
  
  
  Objective 1:
  -------
  
  Whenever primary CARS down completely, standby database becomes a primary database AUTOMATICALLY and it should allow the read/write operation.
  I guess it's called 'Fast Failover'. Please let me know if I'm wrong.
  
  Issues related to the:
  
  -To do this, I need to set up Data Guard Broker so that this standy database becomes primary when CARS go down completely with a power outage, planned or not.
  -Let's say that CARS falls does completely, how long take Data Guard Broker do standby db as primary.
  -What the client application / which is already connected to the CARS.
  -Let's DB standby became as primary and after some time if RAC comes back, keep data automatically becomes the primary role of RAC?
  
  
  Objective 2:
  --------
  
  As I explained above, all products Oracle IDM and applications to speak to the RAC database, what do know only on the RAC database, which is the main. They are not aware of the pending database.
  
  -Whenever a client session is underway with the primary database of CARS if CARS completely falls down, we would like to wait until the client session should get transferred datbase standby without losing session information. However before that happens, standby database should become primary because the client session can perform write operations.
  
  -Whenever a client attempts to connect to the primary CARS and assumes that the cab is completely down, we would like to expect from any client connections should are transferred pending database.
  However before that happens, standby database should become primary because the client session can perform write operations.
  
  According to my knowledge, above scenarios are called "client failver." Please let me know if I'm wrong.
  
  
  
  Issues related to the:
  ----------
  
  1. Please throw some light to reach above features.
  2. According to my understanding, before customer failover happens, fast failover expected has already occurred and ensure should get the switch for the main role. I guess that all this happens thanks to timeout settings. What are those.
  
  
  Could you please help?
  
  Thank you

  I didn't say that clients cannot reconnect automatically when primary fails: I meant that client sessions can generally reconnect and also keep any session state.

  Yes, they can reconnect but only to keep the session state for a SELECT statement. It is also possible with Data Guard, not only in the CARS I wrote: here's a demo with 10 g
  http://uhesse.WordPress.com/2009/08/19/connect-time-failover-transparent-application-failover-for-Data-Guard/.
  What is not possible is to maintain a session state: (INSERT/UPDATE/DELETE) pending transactions must be cancelled.

  Edited by: P. Forstmann on Dec 8. 2011 20:30

• What happens to the application if the node fails at the JRC

  Hello
  
  If an application is connected to the 2 database RAC node and performs some jobs and if it is connected to node 1 and if node 1 fails, it will be resumed at the 2nd node if it is connected thorough a service that is started as favorite on both nodes.
  
  
  If suppose the failover of applications to the 2nd node do we need to submit work again.
  
  Here we are connect to the database from an application not directly.

  If the application is TAF aware it will be. If you select it will be summarized in the other host, if in an application of transaction will trap the failover operation, restore and restart.

  HTH

• What is so special about TAF (compared to the normal recovery)?

  The GI version: 11.2.0.4

  OS: Oracle Linux 6.5

  Question1.

  By default (without FAT), when an instance fails, queries are failed over to another node. So, what advantage a failover Transparent Application does?

  These are documents 11.2 said

  http://docs.Oracle.com/CD/E11882_01/network.112/e10836/advcfg.htm#NETAG338

  "TAF can operate in one of two modes, resumed the Session and select failover." Failover session recreates the sessions and lost connections. Select failover reruns which were ongoing queries. »

  I guess the regular shift (without configured TAF) only Session failover. Am I wrong? Are there other benefits to WORK?

  Question2.

  The doc said below

  http://docs.Oracle.com/CD/E11882_01/Java.112/e16548/ocitaf.htm#JJDBC28836

  "(TAF) transparent Application failover is a feature of the driver Java Database Connectivity (JDBC) Oracle Call Interface (OCI)".

  The connection that we do to the database using development tools GUI like Toad, Allround PL/SQL Developer, SQL Developer use pilot OIC. Right?

  This statement:

  By default (without FAT), when an instance fails, queries are failed over to another node.

  is not correct. There is no failover of any kind unless you implement TAF.

• TAF TNS entry for active data guard

  Hello world

  Need your advice on the configuration of the TNS entrance side transparent client failover of the DB in waiting in a Data Guard configuration active?

  Normally, I create a service manually and configure this service to start or stop according to the change of role of database. Then have personalities of the two sites in the entry of TNS (example below).

  TEST =

  (DESCRIPTION =

  (FAILOVER = ON)

  (ENABLE = BROKEN)

  (LOAD_BALANCE = TRUE)

  (ADDRESS = (PROTOCOL = TCP) (HOST = VIP1-SITE1)(PORT = 1521))

  (ADDRESS = (PROTOCOL = TCP) (HOST = VIP2-SITE1)(PORT = 1521))

  (ADDRESS = (PROTOCOL = TCP) (HOST = VIP1-SITE2)(PORT = 1521))

  (ADDRESS = (PROTOCOL = TCP) (HOST = VIP2-SITE2)(PORT = 1521))

  (CONNECT_DATA =

  (SERVER = DEDICATED)

  (SERVICE_NAME = TEST)

  (FAILOVER_MODE =

  (TYPE = SELECT)

  (METHOD = BASIC)

  (RETRIES = 20)

  (TIME = 2)

  )

  )

  )

  I learned this method of-> http://uhesse.com/2009/08/19/connect-time-failover-transparent-application-failover-for-data-guard/

  Is there a better way to do this? Can we achieve this using an AMT only?

  I think I found the answer. According to the document oracle practices best client failover, the service can be configured to start automatically on the standby server when the role of the db is changed. This can be achieved by creating a service with srvctl with option '-l PRIMARY' on both sites

  For example:

  Main cluster: srvctl add service d Austin s oltpworkload r - ssa1 ssa2 ssa3, ssa4-l primary SCHOOL - q TRUE EI SESSION m BASE w 10 - z 150

  Cluster of relief: srvctl add service d Houston s oltpworkload - r BSR1, ssb2, ssb3, ssb4 - l PRIMARY - q PURE EI SESSION m BASE w 10 - z 150

  Documentary link--> http://www.oracle.com/au/products/database/maa-wp-11gr2-client-failover-173305.pdf

  Also the entrance to the AMT must be created with two descriptions, one for primary and one for standby

  TNS_DG =

  (DESCRIPTION_LIST =

  (LOAD_BALANCE = OFF)

  (FAILOVER = ON)

  (ENABLE = BROKEN)

  (DESCRIPTION =

  (ADDRESS_LIST =

  (LOAD_BALANCE = ON)

  (ADDRESS = (PROTOCOL = TCP) (HOST = vip1-primarydb)(PORT = 1521))

  (ADDRESS = (PROTOCOL = TCP) (HOST = vip2-primarydb)(PORT = 1521))

  )

  (CONNECT_DATA =

  (SERVER = DEDICATED)

  (SERVICE_NAME = orcl)

  (FAILOVER_MODE =

  (TYPE = SELECT)

  (METHOD = BASIC)

  (RETRIES = 20)

  (TIME = 2)

  )

  )

  )

  (DESCRIPTION =

  (ADDRESS_LIST =

  (LOAD_BALANCE = ON)

  (ADDRESS = (PROTOCOL = TCP) (HOST = vip1-standbydb)(PORT = 1521))

  (ADDRESS = (PROTOCOL = TCP) (HOST = vip2-standbydb)(PORT = 1521))

  )

  (CONNECT_DATA =

  (SERVER = DEDICATED)

  (SERVICE_NAME = orcl)

  (FAILOVER_MODE =

  (TYPE = SELECT)

  (METHOD = BASIC)

  (RETRIES = 20)

  (TIME = 2)

  )

  )

  )

  )

• ACFS Filesystem: Register Managed vs managed CRS

  Hello

  Approx.: 11.2.0.4.1 on Solaris. Clusterware only (no Oracle DB). Managed registry ACFS Filesystem.

  I have 3 questions:

  1. can you managed CRS ACFS filesystem intended only in the case of a home DB or even otherwise (for example in the case of another application that is registered to the SIR as a resource and using the ACFS filesystem for read/write and therefore depends on)

  2. If there are open files, a phasing out of the CRS (crs stop crsctl) put an end to these user sessions or hang on? In my case its hanging.

  3. when registry Manbased file system should be used and what should CRS managed filesystem serve?

  Thank you

  Post edited by: user12022918

  The time difference is due to the difference in functionality and installation.  The STOP_TIMEOUT control how long it takes to stop phase complete.  The stop phase will loop continuously try to disassemble, hoping the process will disappear so it can be removed cleanly.  The registry is built in an attempt to deal with more than one system of files for the STOP_TIMEOUT is much greater.  (You can see using crsctl stat res ora.registry.acfs Pei | grep STOP_TIMEOUT.)  Once finished the stop phase, the clean phase may or may not be triggered, according to the release of the stop phase.

  I don't know what you mean by active-passive.  ACFS is designed to run on all nodes in a cluster at any given time.  The scenarios of resource 11.2.0.4 allow not the file system is running on all nodes and allow the application to switch from node to node for active / passive be easily installed - you would be much better to just Setup.  There is no wise sentence for execution ACFS show in a cluster vs. a single node, so the complexity to make active-passive is not worth.  Implementation of the active path, failover of applications would take between 5 and 30 years, the worst case. (basically the time it takes for CRS find that file system is spent in offline and to request the application if it is to be switched.)  "This could do faster by decreasing your check_interval application.)  The resource of the ACFS file system would be the only way to do, because there is no way to support dependencies to install on the resource registry ACFS.  (In addition, the fact that it manage multiple file completely systems would break dependencies, as it might be online w\o the file system, you need to be online.)

• The SYS user connects to the database level, is this correct?

  My senior colleague gave me following information about the user sys. I want to know, - is this correct?
  
  Since the SYS user connects to the database level, therefore, on the session active user SYS, to kill only the current statement is canceled. The database session does not disconnect. Instead, he continues to run the remaining in the script file instructions where we run a script file that contains a large number of SQL statements.
  
  Elya

  If the Oracle session was killed and the customer continued to successfully submit the following commands to the database, a new session was automatically created to replace the killed session. A feature called Of Application Transparent failover which allows the customer to switch to a session of a RAC server to another RAC server, the initial session fails (for example, RAC Server breaks down).

  TAF is part of the response to the so-called "+ the Evil Twins +"-errors and crashes. The rest of the answer are FAN (Fast Application Notification) and CLF (resumption of fast connection). This set of features about the errors and crashes - and allows an application 'seamless' continue even when his Oracle Server session fails.

  See the white paper Oracle on failover of applications with Oracle Database 11 g for the basic details.

• How to find and view the results in flash

  Hello friends,

  I work in Flash 8 and using AS 2.0 coding.

  I work with failover-page application. Here I have an option 'search '. I'll tell you the scenario.

  Here I am loading images/pages dynamically through xml. He loading and works as a page flipping. so far, it has been done. Now if you enter some text inside the search option and press, he must search all loaded images/pages and view pages resulted.

  How this feature is indeed. How to find the jpg with a given string?

  Your help will be really appreciated.

  Kind regards

  Rajesh

  You must have a database of information related to the pages.  If the pages are only jpeg images, then they will not be able to provide you with anything in the way of useful data searchable beyond possibly titles image.  So, you will need to create a database that contains all of the information in these pages that you can have your (matching chain) diagram of function search works with to find matching information.

• CARS @ Oracle University training, advice please

  Greetings,
  
  Can you talk about the environment @ Oracle university for the training of RAC database? They will give training in environment of configuration database RAC or simply on virtual machines.
  
  What is the advantage/difference between Oracle training @ & training in a Training Institute.
  
  
  We are being implemented Oracle 2 node RAC11g db with 2 node application server 10 g (R3) on solaris...
  
  
  
  Thank you very much
  
  Concerning

  I currently teach the Oracle RAC University course.

  There are 4 'normal' ways of this form of courses OR to take:

  -In one place OR using equipment of the Oracle.
  -To a situation of partner OR using equipment of the Oracle.
  -To a situation of partner OR partner devices;
  -At the customer on their equipment.

  The first two grounds involve the participant being in a classroom, computers that run the FCAC to another centre and VNC sessions of the workstation class computers CARS. As long as the network is OK, it's a fantastic way to make the course of the ORGANIZATIONAL unit. (Being a partner of the OU, I am biased toward the second or third.)

  The third has some possibilities.

  -In one, I provide 2 computers with multiple network cards to each student (or team of 2 students), more a FireWire drive double head, or one additional computer that is configured as a server of OpenFiler iSCSI drive. The classroom is equipped with 2 or 3 GB switches. (I use 3 If iSCSI.) Students can install Linux and RAC and clusterware.

  -In another, the Center provides a larger machine, and we have 3 VMs... one like OpenFiler and two as a pre-configured with Oracle Enterprise Linux RAC nodes.

  The fourth is the most interesting, because I end up going in a week before of time (at the responsibility of the customer) and the configuration of their equipment to meet the class structure.

  In all cases, the course material is the same... Conference + lab to install clusterware, ASM, RAC, administration at every level, performance analysis, specific to CARS and tuning show how to implement the failover of applications and early notification of applications and application design from HA and performance.

  A common mistake by customers is to send only DBA. RAC depends on the system, network, and disk directors working with the DBA and they MUST be trained as well.

• rocker and its effect on the program

  Hi dear DBAs,
  
  We have a business application, written using J2EE technologies. It uses a single oracle database. We decided to improve the availability of the application by making a database waiting for our primary database. After some research, we found that datagurad oracle is suited to our need. We have worked with him in a test environment and everything was all right, by using the grid GUI control. The question is, is it enough to set up a database for our main GC pending? or should we change our application code and settings to match the new architecture? In the case of a failover or a failover, how application includes the address of the new database?
  
  Thanks in advance

  Hi Z0lenDer,

  From a database point of view, DataGuard will provide more rapid data recovery and protection of your application data. But for the application to continue working after a failover of the database / failover, you must update your application code and configure to connect to the database from backup when it becomes active.

  This change will refer to the application layer, not in the grid control. You can either use a database trigger or recovery of connection in your application to connect descriptor to ask him to connect to the database instance active.

  Discover the Guide de Configuration DB HA http://download.oracle.com/docs/cd/B19306_01/server.102/b25159/configbp.htm#sthref388

  I hope this helps.

  Farouk

• Failover of the database for applications using JDBC connections

  Hello

  We have DataGuard instances for our Production databases and are configuring failover falls down the primary databases. We are able to set up TNS/LDAP for applications that use this method of connection, however we have a number of applications third-party who connect using JDBC and are not configurable in any other method. Is there something we can do about it or we manually configure these applications if a failover is required?

  Thank you

  Adam

  AdamJ.Sawyer wrote:

  Hello

  We have DataGuard instances for our Production databases and are configuring failover falls down the primary databases. We are able to set up TNS/LDAP for applications that use this method of connection, however we have a number of applications third-party who connect using JDBC and are not configurable in any other method. Is there something we can do about it or we manually configure these applications if a failover is required?

  Thank you

  Adam

  JDBC:Oracle:thin:[user/password]@//[host][:port]/service

  consider that 'HOSTING' can be rather a virtual IP (VIP); which can be configured at the level of the BONE

• How many still for my application to interact with the Timesten database?

  Hi all, still how much for my application to interact with the Timesten database?
  As far as I know, we can connect directly to the database without network load that improves performance, should my application only to specify the DSN name under the name of DSN Manager data? There is no document to clarify that clearly.
  The other approach is c/s connection and there are 3 commucation protocols, such as plain TCP/IP, sockets, UNIX domain and the CPI, right?

  Yes, that's correct. (Preferably) direct mode when the application and TimesTen are on the same machine. UNIX domain socket and shared memory IPC server local (only for cases of special purpose given that the direct mode is faster and more) and normal TCP/IP client/server (local or remote).

  The specific details on how the application connects depends on which API that it uses (ODBC, OCI, Pro * C, JDBC, ODP .NET). For ODBC and JDBC, you must specify a data source name and use the driver corresponding (direct-mode driver or client driver). For the OIC that is configured through SQL * Net and TNSNAMES. ORA and for ODP .NET, I don't remember without it looking up. Please see the TimesTen for relevant API documentation for full details.

  Chris

• failover of transparent applications on the node test system 2 with 9i RAC

  Hi gurus,
  I'm trying to configure transparent applications on my 2 test 9i RAC node failover. 9i RAC (installed due to RAM limitations)
  
  The tnsnames.ora on node1 and node2 are fairly simple with the exception of the IP address.
  ORCL. EZHOME.COM =
  (DESCRIPTION =
  (ADDRESS_LIST =
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.3)(PORT = 1521))
  )
  (CONNECT_DATA =
  (SERVICE_NAME = ORCL)
  )
  )
  
  To test the TAF, I connect to my laptop sqlplus customer who has the following tnsnames.ora
  
  LISTENERS_ORCL. EZHOME.COM =
  (ADDRESS_LIST =
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.3)(PORT = 1521))
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.4)(PORT = 1521))
  )
  
  ORCL. EZHOME.COM =
  (DESCRIPTION =
  (ADDRESS_LIST =
  (LOAD_BALANCE = YES)
  (FAILOVER = ON)
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.3)(PORT = 1521))
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.4)(PORT = 1521))
  )
  (CONNECT_DATA =
  (SERVICE_NAME = ORCL. EZHOME.COM)
  (SERVER = DEDICATED)
  (FAILOVER_MODE =
  (TYPE = SELECT)
  (METHOD = BASIC)
  )
  )
  )
  ORCL1. EZHOME.COM =
  (DESCRIPTION =
  (ADDRESS_LIST =
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.3)(PORT = 1521))
  )
  (CONNECT_DATA =
  (SERVICE_NAME = ORCL. EZHOME.COM)
  (SERVER = DEDICATED)
  (INSTANCE_NAME = ORCL1)
  (FAILOVER_MODE =
  (BACKUP = ORCL2. EZHOME.COM)
  (TYPE = SELECT)
  (METHOD = BASIC)
  )
  )
  )
  ORCL2. EZHOME.COM =
  (DESCRIPTION =
  (ADDRESS_LIST =
  (ADDRESS = (PROTOCOL = TCP) (HOST = 192.168.2.4)(PORT = 1521))
  )
  (CONNECT_DATA =
  (SERVICE_NAME = ORCL. EZHOME.COM)
  (SERVER = DEDICATED)
  (INSTANCE_NAME = ORCL2)
  (FAILOVER_MODE =
  (BACKUP = ORCL1. EZHOME.COM)
  (TYPE = SELECT)
  (METHOD = BASIC)
  )
  )
  )
  
  Select instance_name, host_name,.
  NULL AS failover_type,
  NULL AS failover_method,
  Failed_over NULL AS
  V $ instance
  UNION
  SELECT NULL, NULL, failover_type, failover_method, failed_over
  SESSION $ v
  WHERE username = 'HR ';
  
  RESULT:
  
  INSTANCE_NAME HOST_NAME
  FAILOVER_TYPE FAILOVER_M ISP
  ----------------
  ----------------------------------------------------------------
  ------------- ---------- ---
  orcl1 sunny1pub.ezhome.com
  
  NO NO NO.
  
  2 selected lines.
  
  (1) as you can see, the failover_type, failover_method, failed_over are None, None, and no.. I searched the internet for the last couple days, but I can't find any sample files. Grateful if someone could send me an example of sample tnsnames.ora file.
  
  (2) I need to have the LISTENERS_ORCL. EZHOME.COM? What is the purpose of use / of it?
  
  Help, please. Thank you.

  Hello

  My tnsnames.ora on a client machine to a 10 g RAC, it is similar...

  ORCL =
  (DESCRIPTION =
  (LOAD_BALANCE = ON)
  (FAILOVER = ON)
  (ADDRESS_LIST =
  (ADDRESS = (PROTOCOL = TCP) (HOST = rac1-vip)(PORT = 1521))
  (ADDRESS = (PROTOCOL = TCP) (HOST = rac2-vip)(PORT = 1521))
  )
  (CONNECT_DATA =
  (SERVER = DEDICATED)
  (SERVICE_NAME = ORCL.mycomp.com)
  (FAILOVER_MODE =
  (TYPE = SELECT)
  (METHOD = BASIC)
  (RETRIES = 10)
  (TIME = 1)
  )
  )
  )

  Only this alias nothing more.

  See you soon,.

  Rodrigo Mufalani
  http://mufalani.blogspot.com

  Published by: Mufalani on 02/07/2009 14:19

• FT: No risk for the breakdown of the application during failover?

  When you use a virtual machine which is protected by the functionality of fault tolerance - what happens if the first host fails at a time extremely critical time and the secondary virtual machine is just a few milliseconds after? I think a sort of database or similar server, what happens if something is sent over the network to a client and that action is committed to the customer, and microseconds later the host fails. When the secondary image resumes - is there a risk of application failure or corruption of this?

  Hello.

  To paraphrase it or reducing it to an answer simple will not do justice.  The answer is here.

  Good luck!

• Application is aware of the oracle of failover to standby mode?

  When manually switch you the day before prod, how client application knows to point to the database pending?

  Hello;

  Generally, this is handled with tnsnames.ora and a trigger. So, to answer your question, no, not without help.

  ernie =
   (DESCRIPTION =
      (ADDRESS_LIST =
         (ADDRESS = (PROTOCOL = TCP)(HOST = primary.host)(PORT = 1521))
         (ADDRESS = (PROTOCOL = TCP)(HOST = standby.host)(PORT = 1521))
         )
         (CONNECT_DATA =
         (SERVICE_NAME = ernie)
      )
   )
  

  and then

  CREATE TRIGGER CHECK_ERNIE_START AFTER STARTUP ON DATABASE
  DECLARE
   V_ROLE VARCHAR(30);
  BEGIN
   SELECT DATABASE_ROLE INTO V_ROLE FROM V$DATABASE;
   IF V_ROLE = 'PRIMARY' THEN
   DBMS_SERVICE.START_SERVICE('ernie');
   ELSE
   DBMS_SERVICE.STOP_SERVICE('ernie');
   END IF;
  END;
  /
  

  For more information:

  See DBMS_SERVICE. CREATE_SERVICE

  http://docs.Oracle.com/CD/E11882_01/AppDev.112/e10577/d_serv.htm

  How to switch to a customer in a transparent way in a scenario of switchover of dataguard

  http://www.USN-it.de/index.php/2007/06/28/how-to-fail-over-a-client-transparently-in-a-DataGuard-switchoverfailover-scenario/

  Best regards

  mseberg