[Replicator]
|
Description
|
Factory value
|
Access mode
|
---|---|---|---|
AllowAlreadyStarted
(solidDB version 200.0.2 and later)
|
If set to yes, starting a subscription that is already started does not result in an error.
|
no
|
RW
|
AllowMissingTargetTable
|
If set to yes, CREP ignores missing tables in the target database; replication does not load data into missing tables and does not try to replicate changes to missing tables.
If set to yes and a missing table is then created in the target database, CREP replication does not automatically load the table data from the source. The database administrator should load the data if needed (for example, by using the CREP load command). Note that this might lead to data inconsistency between the databases.
If changed, the new value affects only subscriptions that are started after the change.
|
no
|
RW
|
Appliers
|
Number of applier threads (number of transactions that can be executed by replicator simultaneously).
|
8
|
RW/
Startup |
AutoRestart
|
If set to yes, subscriptions that were ACTIVE at server shutdown are automatically restarted when the server is restarted.
Should be set to no if the server is the target server for a dedicated replication server (to avoid conflicts with dedicated replication process).
|
yes
|
RW/
Startup |
BatchSize
|
Replicator preferred commit block size (in rows) that is used by load.
When load writes data to a target table, the data is not inserted in a single transaction; the load is split into smaller transactions of the specified size.
For example: BatchSize=500
|
200
|
RW/
Startup |
ConcurrentBlobSoftLimit
|
Maximum number of BLOB files that can be read from the source server simultaneously.
|
100
|
RW/
Startup |
ConnectionPoolMaxSize
|
Maximum number of connections in the connection pool.
If this limit is exceeded the replicator does not pool new connections, but drops the connections instead.
|
8
|
RW/
Startup |
DeleteCommitBlock
|
Size of commit block when Replicator is running deletes in load phase while flushing old data. Recommended to be used when Replicator and HotStandby are used together with a large number of rows in some of the replicated tables. The option is unnecessary when Hot Standby is not present. 0 means no limit. The setting value means limit for commit block size. Initial suggestion for value is 10000.
Potential adverse side effect with this setting is loss of performance in the replication load phase.
|
0
|
RW/
Startup |
DurabilityLevel
|
Durability level for replicated transactions.
You can use this parameter to set a different durability level than the server default value (which is set by using Logging.DurabilityLevel).
The set of allowed parameter values is the same as for Logging.DurabilityLevel, see Logging section.
If set to 0, replicated transactions use the server default durability level.
|
0
|
RW
|
ErrorNumberLimit
|
Maximum number of rows that the following system tables can contain before the oldest rows are removed from the table:
▪ SYS_REPLICATION_CONFLICTS
▪ SYS_REPLICATION_ERRORS
|
10000
|
RW/
Startup |
Loaders
|
Maximum number of load reader threads (defines number of tables that can be loaded simultaneously from the same source).
|
10
|
RW/
Startup |
MaxQueueLen
|
Maximum size of Replicator queue length on the target side where the transactions are buffered before execution. Increasing this value will prevent throttling in reading the transactions from source. This will help handling occasional peaks in the load but does not provide help in steady workload replication performance.
|
100
|
RW/
Startup |
MaxRows
(solidDB version 200.0.3 and later)
|
Maximum number of rows that can be sent from the source database to the target database in one network message in CREP replication.
Maximum value is 10000.
|
1000
|
RW/
Startup |
MergeThrottle
|
Enables the replication to check the merge status of the target database. If merge is active, the replication throttles and allows the merge task to proceed faster. This setting will alleviate the interference of replication load with actual workload of the target database.
|
no
|
RW
|
NicePriority
|
(UNIX/Linux operating systems only) Priority of replication threads in the server process.
Valid values are between -20 and 19 inclusive.
Values below 0 increase the priority and values above 0 decrease the priority.
Note On Windows, this parameter is ignored.
|
0
|
RW/
Startup |
NoRefintegrity
|
If set to yes, referential integrity checks are disabled for replicated transactions.
|
yes
|
RW/
Startup |
SerialCommits
|
If set to yes, transactions are committed in the target database in the same order as the transactions were committed in the source database.
If set to no, transactions might be committed in a different order.
Note Transactions that affect the same rows are always committed in the same order in the target database, regardless of the value of this parameter.
|
no
|
RW/
Startup |
SourceIdleTimeout
|
This is the interval in minutes after which the Source Database will drop the connection to Target Database if there has been no responses to ping messages. When the timeout expires, the subscription will go to error state on the Target Database side and needs to be restarted. The connection is dropped on the Source Database side.
|
5
|
RW/
Startup |
SourceLoginTimeout
|
This is the timeout in seconds for the Target Database to log in to the Source Database. When this expires, the connection attempt fails and subscription will go to error state.
|
10
|
RW/
Startup |
SourceReadTimeout
|
Default source connection read timeout (in seconds).
The parameter value can be overridden by specifying a replicator connection timeout option.
The minimum value is 2.
|
30
|
RW/
Startup |