19.2.4.3 Monitoring the Change Stream Applier

The Performance Schema table replication_applier_configuration shows the applier version, worker count, and event-memory limit for each channel:

SELECT CHANNEL_NAME,
       APPLIER_VERSION,
       APPLIER_WORKER_COUNT,
       APPLIER_EVENT_MEMORY_LIMIT
  FROM performance_schema.replication_applier_configuration;

The Performance Schema table replication_applier_status_by_worker shows the state, current transaction, last applied transaction, retries, and most recent error for each CSA worker session:

SELECT CHANNEL_NAME,
       WORKER_ID,
       THREAD_ID,
       SERVICE_STATE,
       APPLYING_TRANSACTION,
       LAST_APPLIED_TRANSACTION,
       LAST_ERROR_NUMBER,
       LAST_ERROR_MESSAGE
  FROM performance_schema.replication_applier_status_by_worker;

After a CSA channel is stopped, its worker rows remain available. THREAD_ID is NULL and SERVICE_STATE is OFF for an inactive worker.

If a CSA job fails, the channel stops and reports an error in the following form:

Replica SQL for channel 'channelName': Change Stream Applier channel stopped
because of error(s): Job 'gtid' failed. See error log or performance schema
tables for more details about this failure or others, if any.
Error_code: code