Configuring the dbclusterndb.gen File

Use the dbclusterndb.gen file to configure the database settings used by each SBR Carrier node to access the SSR database. The dbclusterndb.gen file is stored in the radiusdir directory, usually /opt/JNPRsbr/radius.

[Bootstrap] Section

The [Bootstrap] section (Table 29) of dbclusterndb.gen specifies information that SBR Carrier nodes use to load SSR functions.

Table 29: dbclusterndb.gen [Bootstrap] Fields

Parameter	Function
LibraryName	Specifies the name of the cluster database module. Default value is `dbclusterndb`. Do not change this unless you are advised to do so by Juniper Networks Technical Support.
Enable	If set to 0, the high availability functionality is disabled. If set to 1, the high availability functionality is enabled. Default value is 1 in the file provided with SSR. If this setting is removed from the `dbclusterndb.gen` file, default value switches to 0.
ManagementMode For information about management mode, see the section on Session State Register Administration in the SBR Carrier Administration and Configuration Guide.	If set to 0, the SBR Carrier server operates in standard high-availability mode. If set to 1, the SBR Carrier server operates in management mode. Default value is 0.

Parameter

Function

LibraryName

Specifies the name of the cluster database module. Default value is dbclusterndb.

Do not change this unless you are advised to do so by Juniper Networks Technical Support.

Enable

If set to 0, the high availability functionality is disabled.
If set to 1, the high availability functionality is enabled.
Default value is 1 in the file provided with SSR. If this setting is removed from the dbclusterndb.gen file, default value switches to 0.

ManagementMode

For information about management mode, see the section on Session State Register Administration in the SBR Carrier Administration and Configuration Guide.

If set to 0, the SBR Carrier server operates in standard high-availability mode.
If set to 1, the SBR Carrier server operates in management mode.

Default value is 0.

[NDB] Section

The [NDB] section (Table 30) of dbclusterndb.gen identifies how SBR Carrier nodes access the SSR database.

Table 30: dbclusterndb.gen [NDB] Fields

Parameter	Function
ManagementNode	Specifies the NDB connect-string value, made up of the IP address of the management node hosts, the port the management node uses for connection requests, and the node ID (`NDB connect-string`) of the local SBR Carrier node. This information is created by the installation script and should only be changed by that script or under direction of JTAC.
ConnectRetries	Specifies how many times SBR Carrier tries to connect to the management nodes. Default value is 3.
DelayBetweenConnectRetriesSec	Specifies how many seconds SBR Carrier waits between retries when trying to connect to the management nodes. Default value is 5 seconds.
TimeoutForFirstAliveSec	Specifies how many seconds SBR Carrier waits for the first NDBD server to confirm that it can communicate before communicating with the database cluster. Default value is 10 seconds.
WaitForAllNodesAlive	If set to 0, SBR Carrier does not wait for confirmation that all NDB nodes are alive before communicating with the database cluster. If set to 1, SBR Carrier waits for confirmation that all NDB nodes are alive before communicating with the database cluster. Default value is 0.
TimeoutAfterFirstAliveSec	Specifies how many seconds SBR Carrier waits after the first NDBD server alive indicator. Default value is 10 seconds.
NdbHandles	Specifies the number of NDB handles used for parallel database transactions. One NDB handle is needed for each ongoing database operation. Therefore, the number of available handles poses an upper limit to the number of concurrent operations a SBR Carrier server can make to NDB. Performance increases with concurrency up to a point, where thread overhead overwhelms the benefits of concurrency. The best balance depends on your environment. Note: Each NDB handle uses more than 32K of memory. Increasing the value of NdbHandles increases the start time for SBR Carrier very slightly. Enter a value in the range 1–128. Note that one handle is attached permanently to each cache thread, and one handle is reserved by the system for special usage when in ManagementMode. Note: The NdbHandles setting, and its associated alert message, only counts operational handles, i.e., not those reserved for MgmtMode or for caching threads. Default value is 32.
NdbHandlesAlert	If set to 0, the SBR Carrier server does not record NDB handle messages in its log file. If set to 1, the SBR Carrier server records messages identifying the maximum number of concurrent NDB handles that have been used since the server was restarted in its log file. The messages take the form: `Max concurrent NDB handles =` `N` Default value is 0.
NDBHardErrorThreshold	Specifies the threshold value for NDB hard errors. If the number of hard errors exceeds the threshold value, SBR Carrier starts to monitor the cluster health and determines whether to persist sessions in the local file on the SBR Carrier server on the basis of the value set in the `FallbackLocal` parameter in the `radius.ini` file. You can enter the value in the range from 1 through 100. Default value is 10. This parameter is reloaded every time when SBR Carrier receives a hang-up (HUP) signal.

[Database] Section

The [Database] section (Table 31) of dbclusterdb.gen controls how SBR Carrier front ends accesses the SSR database.

Table 31: dbclusterndb.gen [Database] Fields

Parameter	Function
Database	Specifies the name of the database used by SBR Carrier. Default value is `SteelBeltedRadius`.
ReconnectOnHUP	Controls whether the database cluster is disconnected and reconnected after a HUP signal is sent to the dbcluster plugin. If set to 1, the database reconnects after receiving a HUP signal. If set to 0, the database does not reconnect after receiving a HUP signal. Default value is 0.
Retries	Specifies how many times SBR Carrier tries to connect to the management nodes before giving up. Default value is 6.
DelayBetweenRetriesMillisec	Specifies the base number of milliseconds SBR Carrier waits before retrying a database operation. The first retry delay is 1 x the value specified for DelayBetweenRetries, the second retry delay is 2 x the value specified for DelayBetweenRetries, and the nth retry delay is n x the value specified for DelayBetweenRetries. Default value is 50 milliseconds.
RetryAlertThreshold	Specifies the threshold for recording log messages when a retry that exceeds the threshold is attempted. The log message identifies why the retry was attempted (that is, why the previous attempt failed). If RetryAlertThreshold is set to 0, log messages are written before every database retry. If RetryAlertThreshold is set to the value of (Retries -1), log messages are written before the last retry and after the last retry (if it fails). If RetryAlertThreshold is set to the value of Retries, log messages are written only after the last retry fails (that is, when the last retry failed and no further attempt will be made). If RetryAlertThreshold is set to a value of greater than the value of Retries, log messages are not recorded. Default value is 4294967295 (0xffffffff).
UseConnectionManager	UseConnectionManager=<bool>. This parameter should always be set to true unless a change is recommended by Juniper Networks Technical Support. This might have a minor performance impact on throughput, bandwidth-bound installations. Contact your sales engineer or Juniper Networks Technical Support for more information. The default is enabled.

[IpAddressPools] Section

The [IpAddressPools] section (Table 32) of dbclusterndb.gen specifies aging and caching parameters for IP address pools. Some of these settings are system-wide and cannot be overridden. Other settings establish system defaults, which can be overridden for specific IP address pools.

Table 32: dbclusterndb.gen [IPAddressPools] Fields

Parameter	Function
MinUnusedAgeSec	Specifies how many seconds an IP address can remain unused before SBR Carrier reassigns it. Default value is 300 seconds.
MaxAgeRetries	Specifies the number of times SBR Carrier attempts to retrieve acceptably aged IP addresses before retrieving any available address. If MaxAgeRetries is 0, SBR Carrier makes one attempt to look for addresses that have been idle for at least MinUnusedAgeSec seconds, then no retries if enough addresses are not found. Default value is 2.
AgePercent	Specifies the percentage (a number in the range 0–100) that SBR Carrier uses as the multiplier for MinUnusedAgeSec when enough IP addresses that have been idle for MinUnusedAgeSec cannot be found. For example, if MinUnsedAgeSec is 400 and AgePercent is 75, SBR Carrier would look for addresses idle for at least 400 seconds, then look for addresses that have been idle for at least 300 (400* 0.75) seconds, then look for addresses that have been idle for at least 225 (300 * 0.75) seconds. If AgePercent is 100, only the original MinUnusedAgeSec is used. If AgePercent is 0, then age is disregarded. Default value is 50.
AgeRetryAlertThreshold	Specifies the threshold (0–4294967295) for recording log messages when an aged-based retry that exceeds the threshold is attempted. The log message identifies why the retry was attempted (that is, why the previous attempt failed). If AgeRetryAlertThreshold is set to 0, log messages are written before every age-based retry. If AgeRetryAlertThreshold is set to the value of (MaxAgeRetries -1), log messages are written before the last age-based retry and after the last retry (if it fails). If AgeRetryAlertThreshold is set to the value of MaxAgeRetries, log messages are written only after the last retry fails (that is, when no addresses have been found and no further attempt will be made). If AgeRetryAlertThreshold is set to a value of greater than the value of MaxAgeRetries, log messages are not recorded. Default value is 4294967295 (0xffffffff).
NumCacheThreads	Specifies how many parallel threads SBR Carrier uses to cache IP addresses. Default value is 2.
StartCachingAtBootTime	If set to 1, SBR Carrier fills its IP address pool cache immediately when it is rebooted. If set to 0, SBR Carrier fills its IP address pool cache when it receives an address request. Default value is 1.
CacheThreadSleepMin	Specifies the minimum range of time (0–4294967295) a cache-filling thread waits before it goes to the database to get another cache of IP addresses. Set this parameter to better manage your caching threads. Default value is 1000 milliseconds.
CacheThreadSleepMax	Specifies the maximum range of time (`CacheThreadSleepMin`–4294967295) a cache-filling thread waits before it goes to the database to get another cache of IP addresses. Set this parameter to better manage your caching threads. Default value is 2*`CacheThreadSleepMin` milliseconds.
CacheLowWater	Specifies the minimum number of addresses that must be available in the address cache for an IP address pool. When the number of addresses in a server’s cache falls below the CacheLowWater value, the server begins requesting blocks of IP addresses Default value is 100.
CacheHighWater	Specifies the number of addresses that must be available in a server’s IP address cache for an IP address pool before it stops adding addresses to the cache The CacheHighWater value must be greater than or equal to the CacheLowWater value. Default value is 250.
CacheChunkSize	Specifies the number of addresses to retrieve every time SBR Carrier requests a block of IP addresses for an IP address pool. Default value is 50.
EmergencyChunkSize	Specifies the (0–`CacheChunkSize`) number of addresses to retrieve every time SBR Carrier requests a block of IP addresses to use from a pool, when that pool’s cache is empty, to directly retrieve those addresses from the IP address table in the database and put them in the cache. Default value is 1.
CacheAlertThreshold	Specifies the threshold (0–4294967295) for recording log messages when the number of addresses in the cache falls below the threshold value. If CacheAlertThreshold is set to 0, log messages are not written. If CacheAlertThreshold is set to 1, log messages are written when the address cache is empty. If CacheAlertThreshold is set to a value equal to or greater than the sum of the values of CacheHighWater and CacheChunkSize, log messages are written whenever an address is pulled from the cache. Default value is 0.
CacheThreadVerbose	If set to 1, print out an informational message about what the caching thread just did. The message contains: thread identity (OS thread ID number), how long it napped, pool cached, number of IP addresses cached, and the length of time it took to retrieve the addresses from the database. If set to 0, no caching thread information is recorded. Use this parameter to fine-tune the caching parameters. Default value is 0.

[IpAddressPools:PoolName] Section

The [IpAddressPools:PoolName] section (Table 33) of dbclusterdb.gen identifies how SBR Carrier specifies the override settings for IP address aging for a named IP address pool. You can create as many [IpAddressPools:PoolName] section as you require in the dbclusterdb.gen file to tune caching for individual address pools.

Note

You cannot use an [IpAddressPools:PoolName] section to override settings other than those listed here. For example, you cannot enter a MaxAgeRetries setting in an [IpAddressPools:PoolName] section to override the value specified in the [IpAddressPools] section.

Table 33: dbclusterndb.gen [IpAddressPools:PoolName] Fields

Parameter	Function
StartCachingAtBootTime	If set to 1, SBR Carrier fills its IP address pool cache immediately when it is rebooted. If set to 0, SBR Carrier fills its IP address pool cache when it receives an address request. Default value is 0.
CacheLowWater	Specifies the minimum number of addresses that must be available in the named pool’s IP address cache. When the number of addresses in the pool’s cache falls below the CacheLowWater value, the server begins requesting blocks of IP addresses.
CacheHighWater	Specifies the number of addresses that must be available in a server’s IP address cache before it adding addresses to the IP address cache for the named IP address pool. The CacheHighWater value must be greater than or equal to the CacheLowWater value.
CacheChunkSize	Specifies the number of addresses to retrieve every time SBR Carrier requests a block of IP addresses for the named IP address pool.
EmergencyChunkSize	Specifies the (0–`CacheChunkSize`) number of addresses to retrieve every time SBR Carrier requests a block of IP addresses to use from a pool, when that pool’s cache is empty, to directly retrieve those addresses from the IP address table in the database and put them in the cache. Default value is 1.
CacheAlertThreshold	Specifies the threshold (0–4294967295) for recording log messages when the number of addresses in the cache falls below the threshold value. If CacheAlertThreshold is set to 0, log messages are not written. If CacheAlertThreshold is set to 1, log messages are written when the address cache is empty. If CacheAlertThreshold is set to a value equal to or greater than the sum of the values of CacheHighWater and CacheChunkSize, log messages are written whenever an address is pulled from the cache. Default value is 0.