Debezium on oracle - CDB vs non CDB

[Prasad]

Hi,

Is there any known issue with running Debezium connectors for oracle on CDB (container DB)? 

We are observing that incremental changes (schema_only mode) works fine on non CDB Oracle. But on CDB there are issues. 

1. At times sluggishness is observed. A change done reflects after more than 30 mins.

It used to work instantly earlier.

2. Incremental changes are not reflecting at all. 

As part of the connector configuration we are using almost same except the PDB parameter for container DB. 

Appreciate your help.

Thanks,

-Prasad  

[Chris Cranford]

Hi Prasad -

There are no issues that we are aware, in fact our default test environment is CDB as that's Oracle's installation default. 

Slowness is typically indicative of a configuration problem.  I would first check the size of the BYTES column in V$LOG versus the configured "log.mining.strategy".  The data dictionary process step of LogMiner can be slow, particularly if you have more than one PDB in your environment, so I would suggest making sure that you have a reasonable BYTES size set for your redo logs to minimize frequent redo log switches as the dictionary will need to be gathered and parsed on each switch.  Another option is to use "log.mining.strategy=online_catalog", if you are not, to avoid this data dictionary write and parse step and to use the current data dictionary as reference.

If you're not seeing changes, this can also be indicative that supplemental logging is not properly configured or that you could be using a tablespace that is intentionally excluded from capture.

Could you please share your full connector configuration and can you also verify that you followed the proper configuration for the users as outlined in the documentation [1].

Thanks

[1]: https://debezium.io/documentation/reference/2.5/connectors/oracle.html#setting-up-oracle

--
You received this message because you are subscribed to the Google Groups "debezium" group.
To unsubscribe from this group and stop receiving emails from it, send an email to debezium+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/debezium/3d512b4d-9e69-4772-881f-a9be74242eaen%40googlegroups.com.

[Prasad]

Thanks Chris. 

I am getting the points validated from DBA. 
Sharing the connector details below

{
"name":"capture_test_ora_init",
"config":{
"connector.class": "io.debezium.connector.oracle.OracleConnector",
"database.hostname": "XXXXXX",
"database.port": "1521",
"database.dbname": "ADVCDB",
"database.pdb.name": "INSESSST",
"database.user": "C##DBZUSER",
"database.password": "XXXXXX",
"tasks.max": "1",
"snapshot.max.threads": "4",
"table.include.list": "<schema_nm>.<table1>",
"topic.prefix": "orainit",
"transforms": "route",
"transforms.route.type": "org.apache.kafka.connect.transforms.RegexRouter",
"transforms.route.regex": "([^.]+)\\.([^.]+)\\.([^.]+)",
"transforms.route.replacement": "$1.$3",
"decimal.handling.mode": "double",
"poll.interval.ms": "30",
"time.precision.mode": "connect",
"schema.include.list": "<schema_nm>",
"schema.history.internal.kafka.bootstrap.servers": "kafka-1-0.kafka-1.debezium.cluster.local:9092",
"schema.history.internal.kafka.topic": "schema-changes.orainit",
"schema.history.internal.store.only.captured.tables.ddl": "true",
"key.converter": "io.confluent.connect.avro.AvroConverter",
"value.converter": "io.confluent.connect.avro.AvroConverter",
"key.converter.schema.registry.url": "http://schemaregistry-0.schemaregistry.debezium.cluster.local:8081",
"value.converter.schema.registry.url": "http://schemaregistry-0.schemaregistry.debezium.cluster.local:8081",
"snapshot.mode": "schema_only"

  }
}

[Prasad]

Update: We started seeing the lethargy in non cdb environment too. But after re-configuring supplemental logging and forced logging changes are reflecting. Had used heartbeat parameter earlier which was not working. But it is working in non CDB. For CDB environment though it was confirmed by DBA that the required configuration is in place, we are now setting up new CDB and verify the behavior. 

Thanks,

-Prasad 

[Chris Cranford]

Sounds good Prasad, let us know how it goes.  If you still see any lethargy, could you please provide as much detail as to how the PDB is set-up, including log groups and what exactly is supplementally logged, etc.

Thanks,
Chris

To view this discussion on the web visit https://groups.google.com/d/msgid/debezium/da909d79-216b-4c29-bf0c-f86ebd6e6a55n%40googlegroups.com.