Mining properties

Generated from the connector's ConfigDef by ConfigDocsGeneratorTest; do not edit by hand.

Property	Type	Default	Importance	Description
cdc.mining.target.latency.ms	long	2000	medium	Latency goal for the adaptive mining window. The only tuning knob: window size follows from it.
cdc.mining.max.logs.per.step	int	8	low	Upper bound on redo logs mined in one step while catching up.
cdc.mining.fetch.size	int	10000	low	JDBC fetch size for V$LOGMNR_CONTENTS.
cdc.mining.query.timeout.ms	long	600000	low	Per-step timeout. On expiry the step is discarded, the session restarted and the window halved; it never fails the task by itself. A socket read on any database connection times out one minute after this, so a connection the network drops silently is reopened rather than waited on.
cdc.mining.session.max.age.ms	long	3600000	low	The LogMiner session is restarted at this age to release PGA.
cdc.mining.decode.threads	int	0	low	Decode threads; 0 means the number of cores minus one, at most 8.
cdc.mining.inlist.max	int	1000	low	Most captured object ids in one IN list of the mining query; more ids are split across several IN lists. Oracle allows at most 1000 entries in one list, and the task fails at start with a larger value.
cdc.mining.catchup.threshold.ms	long	300000	low	Reserved for the lag that would enable parallel catch-up mining, which is not built yet.
cdc.mining.catchup.parallelism	int	2	low	Reserved for the number of catch-up sessions over adjacent SCN windows. Parallel catch-up mining is not built yet.
cdc.rac.safety.lag.ms	long	-1	low	On a database with more than one enabled redo thread (RAC), how far behind the current SCN the online safe end is held: each step mines only to an SCN read at least this many milliseconds earlier, so a slow instance has written its redo below it. -1 means 3000 on such a database. Ignored with one enabled redo thread, where the safe end is the current SCN, and in archive-only mode. RAC capture is refused at start until it is qualified.
cdc.dictionary.build.interval.ms	long	86400000	low	Interval between data dictionary builds into the redo (DBMS_LOGMNR_D.BUILD), which let the connector decode rows written before a later DDL on their table. Needs EXECUTE ON DBMS_LOGMNR_D; without it builds are switched off with an ops event. One build also runs at start when the archived logs hold none. 0 switches builds off.
cdc.dictionary.build.time	string	02:00	low	Time of day, in the database's time, of the first scheduled dictionary build; later builds follow every cdc.dictionary.build.interval.ms.
cdc.dictionary.database.url	string	none	low	JDBC URL of the database dictionary builds run on, with the connector's user and password. Set it to the primary when the connector captures from a physical standby, which is read-only and cannot write a build; the build reaches the standby through the redo. Empty: the captured database.