Oracle® TimesTen In-Memory Database Replication Guide 11g Release 2 (11.2.2) E21635-11 |
|
|
View PDF |
The attributes defined in this chapter are used to set up TimesTen active standby pairs that are managed by Oracle Clusterware. These attributes are specified in the cluster.oracle.ini
file. The ttCWAdmin
utility creates and administers active standby pairs based on the information in the cluster.oracle.ini
file.
This section lists the TimesTen configuration attributes for Oracle Clusterware in these tables:
Name | Description | Default |
---|---|---|
Lists host names that may contain master databases in an active standby pair scheme. |
None |
Table A-2 Conditionally required attributes
Name | Description | Default |
---|---|---|
Command line for checking the status of a TimesTen application that is managed by Oracle Clusterware. |
None |
|
The name of a TimesTen application that is managed by Oracle Clusterware. |
None |
|
Command line for starting a TimesTen application that is managed by Oracle Clusterware. |
None |
|
Command line for stopping a TimesTen application that is managed by Oracle Clusterware. |
None |
|
The database to which the application should link. |
None |
|
Specifies whether the active standby pair replicates cache groups. |
N |
|
Lists the port numbers used by the cache grid agents for the active database and the standby database in an active standby pair that is a cache grid member. |
None |
|
A list of two virtual IP addresses that can be associated with the master databases. |
None |
|
A list of subscriber hosts that are not part of the cluster. |
None |
|
The directory to which the active database is backed up. |
None |
|
List of host names that can contain subscriber databases. |
None |
|
The list of virtual IP addresses that can be associated with subscriber databases. |
None |
|
The name of the public network adapter that is to be used for virtual IP addresses on each host. |
None |
|
The netmask of the virtual IP addresses. |
None |
Name | Description | Default |
---|---|---|
The number of seconds that the Oracle Clusterware resource that monitors the application waits after a failure is detected before performing a failover. |
0 |
|
The interval in seconds before which Oracle Clusterware stops a TimesTen application if the application has exceeded the number of failures specified by the Oracle Clusterware |
60 |
|
The number of consecutive Oracle Clusterware resource failures that Oracle Clusterware tolerates for the action script for an application within an interval equal to 10 * AppScriptTimeout. The default is 2. |
2 |
|
The number of times that Oracle Clusterware attempts to restart the TimesTen application on the current host before moving the application. |
100 |
|
The number of seconds the TimesTen application container waits for the action scripts to complete for a specific application. |
60 |
|
The number of seconds that a TimesTen application must be up before Oracle Clusterware considers the application to be stable. If the application fails within this threshold, the failure is considered a failure to start properly, and the application is restarted on another host. If the application is active past this threshold, it is considered stable. If it fails after being stable, the application is restarted on the same host. |
600 |
|
Specifies whether an active database should be automatically recovered from a backup if both master databases fail. |
No |
|
|
Defaults to |
Active |
The number of seconds that Oracle Clusterware waits before migrating a database to a new host after a failure. |
60 |
|
The number of consecutive failures of resources managed by Oracle Clusterware that are tolerated within 10 seconds before the active standby pair is considered failed and a new active standby pair is created on spare hosts using the automated backup. |
2 |
|
GridIPConf |
Identifies what IP addresses to use for grid:
|
Private |
A list of all desired replication scheme |
None |
|
The number of seconds between each backup of the active database. |
0 (disabled) |
|
A SQL construct of the active standby pair scheme. |
None |
|
The number times an incremental backup occurs between full backups. |
5 |
|
The return service attribute of the active standby pair scheme. |
None |
|
The list of all desired replication scheme |
None |
|
The number of seconds that Oracle Clusterware waits for the monitor process to start before assuming a failure. |
1209600 seconds, or 14 days |
These attributes must be present for each DSN in the cluster.oracle.ini
file. They have no default values.
The required attributes are listed in Table A-1, "Required attributes" and described in detail in this section.
This attribute lists the host names that can contain master databases in the active standby pair. The first host listed has the active master database and the second host has the standby master database when the cluster is initially started. The following are exceptions to this designated order:
If there are already active and standby master databases on specific nodes when the cluster is stopped, then the active and standby master databases remain on those hosts when the cluster is restarted.
If the cluster is started and the only existing database is on a host that is not listed first in MasterHosts
, then that host is to be configured with the active master database. The first host listed for MasterHosts
is to be the standby master database.
You can specify more than two hosts as master hosts for advanced availablity. See "Configuring advanced availability" for details.
If you have only two master hosts configured, you should not define any virtual IP addresses. If more than two master hosts are listed, you should configure virtual IP addresses for internal TimesTen processes to manage replication in the situation of an unrecoverable failure of one of the hosts on which an active or standby database resides. See "Configuring advanced availability" for details.
Set MasterHosts
as follows:
How the attribute is represented | Setting |
---|---|
MasterHosts |
A comma-separated list of host names. The first host listed becomes the initial active master in the active standby pair. The second host listed becomes the initial standby master in the active standby pair.
The hosts defined must be different than the hosts defined for the |
These attributes may be required depending on the desired Oracle Clusterware configuration. They have no default values. The conditionally required attributes are listed in Table A-2, "Conditionally required attributes" and described in detail in this section.
This attribute specifies the full command line for executing a user-supplied script or program that checks the status of the TimesTen application specified by AppName
. It must include the full path name of the executable. If there are spaces in the path name, enclose the path name in double quotes.
The command should be written to return 0
when the application is running and a nonzero number when the application is not running. When Oracle Clusterware detects a nonzero value, it takes action to recover the failed application.
Set AppCheckCmd
as follows:
How the attribute is represented | Setting |
---|---|
AppCheckCmd |
A string representing the command line for executing an application that checks the status of the application specified by AppName . |
This attribute sets the interval in seconds before which Oracle Clusterware stops a TimesTen application if the application has exceeded the number of failures specified by the Oracle Clusterware FAILURE_THRESHOLD
resource attribute. If the value is zero, then failure tracking is disabled.
For more information about the Oracle Clusterware FAILURE_THRESHOLD
resource attribute, see Oracle Clusterware Administration and Deployment Guide.
This attribute specifies the name of a TimesTen application managed by Oracle Clusterware. Oracle Clusterware uses the application name to name the corresponding resource. Any description of an application in the cluster.oracle.ini
file must begin with this attribute.
This attribute specifies the number of times that Oracle Clusterware attempts to restart the TimesTen application on the current host before moving the application to another host.
This attribute specifies the command line that starts the TimesTen application specified by AppName
. It must include the full path name of the executable. If there are spaces in the path name, enclose the path name in double quotes.
Set AppStartCmd
as follows:
How the attribute is represented | Setting |
---|---|
AppStartCmd |
A string that represents the command line for starting the application specified by AppName . |
This attribute specifies the command line that stops the TimesTen application specified by AppName
. It must include the full path name of the executable. If there are spaces in the path name, enclose the path name in double quotes.
Set AppStopCmd
as follows:
How the attribute is represented | Setting |
---|---|
AppStopCmd |
A string that represents the command line for stopping the application specified by AppName . |
This attribute determines the hosts on which the TimesTen application should start.
Set AppType
as follows:
How the attribute is represented | Setting |
---|---|
AppType |
Active - The application starts on the active database of an active standby pair.
|
This attribute specifies the value for the Oracle Clusterware UPTIME_THRESHOLD
resource attribute.The value represents the number of seconds that a TimesTen application must be up before Oracle Clusterware considers the application to be stable.
For more information about UPTIME_THRESHOLD
, see Oracle Clusterware Administration and Deployment Guide.
How the attribute is represented | Setting |
---|---|
AppUptimeThreshold |
Number of seconds. The default is 600. For example:
AppUptimeThreshold=60 |
If the active standby pair replicates cache groups, set this attribute to Y
. If you specify Y
, Oracle Clusterware assumes that TimesTen is connected to an Oracle database and prompts for the Oracle database password.
This attribute lists the port numbers used by the cache grid agents for the active database and the standby database in an active standby pair that is a cache grid member. The port numbers are separated by a comma. This is a mandatory parameter when global cache groups are present.
This attribute defines two virtual IP (VIP) addresses associated with the two master hosts that are used by the active standby pair. If you configure more than two master hosts for advanced availability, you must also define two VIP addresses. However, these VIP addresses are used solely by internal TimesTen processes for managing the active standby pair in the case of an unrecoverable error on one of the master hosts. Use of these VIP addresses by any user or application can result in an error. See "Configuring advanced availability" for details.
This attribute contains a list of subscriber hosts that are part of the active standby pair replication scheme but are not managed by Oracle Clusterware.
This attribute indicates the directory where the backup of the active database is stored. This must be a directory in a shared file system that every node in the cluster can access. This attribute is required only if RepBackupPeriod
is set to a value other than 0.
The directory must be shared by all hosts in the cluster. The shared storage must be NFS or OCFS (Oracle Cluster File System).
If you want to enable backup, install OCFS on the shared storage during the Oracle Clusterware installation process. You can use this shared storage for backup for an active standby pair.
See "Recovering from permanent failure of both master nodes" and "Failure and recovery for active standby pair grid members" for restrictions on backups.
Lists the host names that can contain subscriber databases. If the active standby pair is configured with subscribers, this attribute is required. It has no default value. You should have at least as many host names listed as there are subscribers defined.
If you are using advanced availability for your subscribers, define additional hosts that can be instantiated to take the place of any subscriber host that may encounter an unrecoverable error. Also, when using advanced availability, configure VIP addresses for every current subscriber in use. For example, if you have three subscribers and two additional hosts that can be used for advanced availability, you should have three VIP addresses defined. See SubscriberVIP
for more information.
Set SubscriberHosts
as follows:
How the attribute is represented | Setting |
---|---|
SubscriberHosts |
A comma-separated list of host names.
The hosts defined must be different than the hosts defined for the If not using advanced availability, the order of the hosts defined determines which application with an When using advanced availability, the first hosts listed are used for the subscriber hosts. The latter hosts defined are used for advanced availability. |
This attribute configures a list of the virtual IP (VIP) addresses associated with subscriber hosts that are used for advanced availability. When you configure more hosts than there are subscribers (for advanced availability), you must also define VIP addresses (one for each subscriber). However, these VIP addresses are used solely by internal TimesTen processes for managing replication in the case of an unrecoverable error on one of the subscriber hosts. Use of these VIP addresses by any user or application can result in an error. See "Configuring advanced availability" for details.
Set SubscriberVIP
as follows:
How the attribute is represented | Setting |
---|---|
SubscriberVIP |
One or more virtual IP addresses. You should define the same number of VIP addresses as the number of subscribers.
The order of subscriber virtual IP addresses is used to determine which application with an |
These attributes are optional and have no default values. The optional attributes are listed in Table A-3, "Optional attributes" and described in detail in this section.
This attribute specifies the number of seconds that the process that is monitoring the application waits after a failure is detected before performing a failover. The default is 0.
This attribute specifies the number of consecutive failures that Oracle Clusterware tolerates for the action script for an application within an interval equal to 10 * AppScriptTimeout
. The default is 2.
This attribute indicates the number of seconds that the TimesTen application monitor process waits for the start action script and the stop action script to complete for a specific application. The check action script has a nonconfigurable timeout of five seconds and is not affected by this attribute.
Specifies whether Oracle Clusterware automatically recovers the active database from the backup in the case of a failure of both masters.
If recovery is not automated (AutoRecover
=N
), the database can be recovered using the ttCWAdmin
-restore
command.
You cannot use AutoRecover
if you are using cache groups in your configuration or if a cache grid is configured.
This attributes specifies the number of seconds that Oracle Clusterware waits before migrating a database to a new host after a failure. Oracle Clusterware does not relocate a database if the database comes up during the delay period. This is applicable when advanced availability is configured. The default is 60 seconds.
This attribute specifies the number of consecutive failures of resources managed by Oracle Clusterware that are tolerated within 10 seconds before the active standby pair is considered failed and a new active standby pair is created on spare hosts using the automated backup. A spare node is only an option when using virtual IP addresses.
Oracle Clusterware tries to perform a duplicate for the active standby pair when a single failure occurs; it tries to perform a restoration if more than a single failure occurs.
This value is ignored for basic availability, since a spare node is only configured when at least one virtual IP address is configured, or is ignored when RepBackupPeriod
is set to 0 when using advanced availability, which does include the configuration of at least one virtual IP address.
Note:
TimesTen tolerates only one failure of a backup resource, regardless of the setting for this attribute.Set FailureThreshold
as follows:
How the attribute is represented | Setting |
---|---|
FailureThreshold |
An integer representing the number of consecutive failures of resources managed by Oracle Clusterware that are tolerated within 10 seconds before the active standby pair is considered failed and a new active standby pair is created on spare hosts using the automated backup. The default is 2. |
This attribute indicates the desired replication scheme STORE
attributes for the master databases. The STORE
attributes apply to both the active and standby databases. For more information on STORE
attributes, see "Setting STORE attributes".
This attribute is not required when RepDDL
is configured.
If this attribute is not set, the STORE
attributes take their default values.
This attribute indicates the number of seconds between each backup of the active database. If this attribute is set to a value greater than 0, you must also specify a backup directory by setting RepBackupDir
.
See "Recovering from permanent failure of both master nodes" and "Failure and recovery for active standby pair grid members" for restrictions on backups.
This attribute represents the SQL statement that creates the active standby pair. Use this attribute only in special circumstances. For example, you must specify RepDDL
if you need to exclude tables and sequences from the active standby pair.
If RepDDL
is set, do not set these attributes:
Replace the database file name prefix in the SQL statement with the <DSN>
macro. Use the <MASTERHOST[1]>
, <MASTERHOST[2]>
and <SUBSCRIBERHOST[
n
]>
macros instead of the host names.
There is no default value for RepDDL
.
This example sets RepDDL
for two master databases:
RepDDL=CREATE ACTIVE STANDBY PAIR <DSN> ON <MASTERHOST[1]>, <DSN> ON <MASTERHOST[2]>
See "Using the RepDDL attribute" for additional examples.
You do not usually need to set the ROUTE
clause in RepDDL
because the transmitter of the replication agent automatically obtains the private and public network interfaces that Oracle Clusterware uses. However, if hosts have network connectivity for replication schemes that are not managed by Oracle Clusterware, then RepDDL
needs to include the ROUTE
clause.
If this attribute is used, each STORE
clause must be followed by the pseudo host names such as:
ActiveHost
ActiveVIP
StandbyHost
StandbyVIP
SubscriberHost
SubscriberVIP
This attribute specifies the number of incremental backups between full backups. The number of incremental backups depends on the capacity of the shared storage.
Setting this attribute can impact performance. There is a trade-off between the storage capacity and the time consumption for backup. An incremental backup can be performed much faster than a full backup. However, storage consumption increases until a full backup is performed.
See "Recovering from permanent failure of both master nodes" and "Failure and recovery for active standby pair grid members" for restrictions on backups.
This attribute specifies the return service for the active standby replication scheme. See "Using a return service".
If no value is specified for this attribute, the active standby pair is configured with no return service.
This attribute indicates the replication scheme STORE
attributes of subscriber databases. The STORE
attributes apply to all subscribers. For more information on STORE
attributes, see "Setting STORE attributes".
This attribute is not required when RepDDL
is present.
If this attribute is not set, the STORE
attributes take their default values.
This attribute specifies the number of seconds that Oracle Clusterware waits for the monitor process to start before assuming a failure.
Oracle TimesTen recommends setting a value of several hours because the action script may take a long time to duplicate the active database. The default is 1209600 seconds (14 days).