ERROR Failed due to error: Aborting snapshot due to error when last running 'UNLOCK TABLES': Only REPEATABLE READ isolation level is supported for START TRANSACTION WITH CONSISTENT SNAPSHOT in RocksDB Storage Engine. [DBZ-1072]

Migrated from [DBZ-1072](https://issues.redhat.com/browse/DBZ-1072)

{code:java}
[2019-01-09 14:12:53,114] INFO rpl_skip_tx_api = OFF (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,114] INFO session_track_gtids = OFF (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO slave_gtid_info = OFF (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO slave_type_conversions = (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO sync_binlog = 1000 (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO system_time_zone = CST (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO time_zone = SYSTEM (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO tls_version = TLSv1,TLSv1.1,TLSv1.2 (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO tokudb_version = 7.5.6 (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO tx_isolation = READ-COMMITTED (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO tx_read_only = OFF (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO version = 5.6.16-log (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO version_comment = Source distribution (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO version_compile_compiler = GNU (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,115] INFO version_compile_compiler_major = 4 (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,116] INFO version_compile_compiler_minor = 8 (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,116] INFO version_compile_machine = x86_64 (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,116] INFO version_compile_os = Linux (io.debezium.connector.mysql.SnapshotReader:801)
[2019-01-09 14:12:53,116] INFO Step 0: disabling autocommit and enabling repeatable read transactions (io.debezium.connector.mysql.SnapshotReader:243)
[2019-01-09 14:12:53,228] INFO Step 1: flush and obtain global read lock to prevent writes to database (io.debezium.connector.mysql.SnapshotReader:265)
[2019-01-09 14:12:53,302] INFO Step 2: start transaction with consistent snapshot (io.debezium.connector.mysql.SnapshotReader:284)
[2019-01-09 14:12:53,379] INFO Step 1: releasing global read lock to enable MySQL writes (io.debezium.connector.mysql.SnapshotReader:658)
[2019-01-09 14:12:53,459] INFO Writes to MySQL tables prevented for a total of 00:00:00.157 (io.debezium.connector.mysql.SnapshotReader:668)
[2019-01-09 14:12:53,460] ERROR Failed due to error: Aborting snapshot due to error when last running 'UNLOCK TABLES': Only REPEATABLE READ isolation level is supported for START TRANSACTION WITH CONSISTENT SNAPSHOT in RocksDB Storage Engine. (io.debezium.connector.mysql.SnapshotReader:187)
org.apache.kafka.connect.errors.ConnectException: Only REPEATABLE READ isolation level is supported for START TRANSACTION WITH CONSISTENT SNAPSHOT in RocksDB Storage Engine. Error code: 3045; SQLSTATE: HY000.
at io.debezium.connector.mysql.AbstractReader.wrap(AbstractReader.java:208)
at io.debezium.connector.mysql.AbstractReader.failed(AbstractReader.java:186)
at io.debezium.connector.mysql.SnapshotReader.execute(SnapshotReader.java:720)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
Caused by: java.sql.SQLException: Only REPEATABLE READ isolation level is supported for START TRANSACTION WITH CONSISTENT SNAPSHOT in RocksDB Storage Engine.
at com.mysql.cj.jdbc.exceptions.SQLError.createSQLException(SQLError.java:129)
at com.mysql.cj.jdbc.exceptions.SQLError.createSQLException(SQLError.java:97)
at com.mysql.cj.jdbc.exceptions.SQLExceptionsMapping.translateException(SQLExceptionsMapping.java:122)
at com.mysql.cj.jdbc.StatementImpl.executeInternal(StatementImpl.java:782)
at com.mysql.cj.jdbc.StatementImpl.execute(StatementImpl.java:666)
at io.debezium.jdbc.JdbcConnection.executeWithoutCommitting(JdbcConnection.java:1071)
at io.debezium.connector.mysql.SnapshotReader.execute(SnapshotReader.java:286)
... 3 more
[2019-01-09 14:12:53,525] INFO WorkerSourceTask{id=nengyuanyun-0} Committing offsets (org.apache.kafka.connect.runtime.WorkerSourceTask:328)
[2019-01-09 14:12:53,525] INFO WorkerSourceTask{id=nengyuanyun-0} flushing 0 outstanding messages for offset commit (org.apache.kafka.connect.runtime.WorkerSourceTask:345)
[2019-01-09 14:12:53,526] ERROR WorkerSourceTask{id=nengyuanyun-0} Task threw an uncaught and unrecoverable exception (org.apache.kafka.connect.runtime.WorkerTask:172)
org.apache.kafka.connect.errors.ConnectException: Only REPEATABLE READ isolation level is supported for START TRANSACTION WITH CONSISTENT SNAPSHOT in RocksDB Storage Engine. Error code: 3045; SQLSTATE: HY000.
at io.debezium.connector.mysql.AbstractReader.wrap(AbstractReader.java:208)
at io.debezium.connector.mysql.AbstractReader.failed(AbstractReader.java:186)
at io.debezium.connector.mysql.SnapshotReader.execute(SnapshotReader.java:720)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
Caused by: java.sql.SQLException: Only REPEATABLE READ isolation level is supported for START TRANSACTION WITH CONSISTENT SNAPSHOT in RocksDB Storage Engine.
at com.mysql.cj.jdbc.exceptions.SQLError.createSQLException(SQLError.java:129)
at com.mysql.cj.jdbc.exceptions.SQLError.createSQLException(SQLError.java:97)
at com.mysql.cj.jdbc.exceptions.SQLExceptionsMapping.translateException(SQLExceptionsMapping.java:122)
at com.mysql.cj.jdbc.StatementImpl.executeInternal(StatementImpl.java:782)
at com.mysql.cj.jdbc.StatementImpl.execute(StatementImpl.java:666)
at io.debezium.jdbc.JdbcConnection.executeWithoutCommitting(JdbcConnection.java:1071)
at io.debezium.connector.mysql.SnapshotReader.execute(SnapshotReader.java:286)
... 3 more
[2019-01-09 14:12:53,526] ERROR WorkerSourceTask{id=nengyuanyun-0} Task is being killed and will not recover until manually restarted (org.apache.kafka.connect.runtime.WorkerTask:173)
[2019-01-09 14:12:53,526] INFO [Producer clientId=producer-13] Closing the Kafka producer with timeoutMillis = 30000 ms. (org.apache.kafka.clients.producer.KafkaProducer:1067)
[2019-01-09 14:13:52,014] INFO WorkerSourceTask{id=nengyuanyun-0} Committing offsets (org.apache.kafka.connect.runtime.WorkerSourceTask:328)
{code}

Start in io.debezium.connector.mysql.SnapshotReader, especially the snapshot execution around lines 243-286 and the failure path near line 720; inspect how the transaction isolation level is set before START TRANSACTION WITH CONSISTENT SNAPSHOT. Trace the JDBC execution through io.debezium.jdbc.JdbcConnection.executeWithoutCommitting. Done means a snapshot using the reported RocksDB configuration completes without the SQLSTATE HY000 error.

Written by the indexing model from the issue text.