mac osx 10.12.5

Can you provide some additional informations :

• Server and gerrit version
• connection string
• if possible, the full stacktrace.

No need to other informations, i found the issue.
Gerrit use https://gerrit.googlesource.com/gwtorm internally.
When executing batch, gerrit verify if update result is has expected : https://gerrit.googlesource.com/gwtorm/+/a89d98dbcc1ac117ef0f6fb4c2a3ec875d325b1b/src/main/java/com/google/gwtorm/jdbc/JdbcAccess.java#456
update results is set by https://gerrit.googlesource.com/gwtorm/+/a89d98dbcc1ac117ef0f6fb4c2a3ec875d325b1b/src/main/java/com/google/gwtorm/schema/sql/SqlDialect.java#435

Has JDBC executeBatch required, the int[] result can contain :

• A number greater than or equal to zero – indicates that the command was processed successfully and is an update count giving the number of rows in the database that were affected by the command's execution
• A value of SUCCESS_NO_INFO – indicates that the command was processed successfully but that the number of rows affected is unknown
• A value of EXECUTE_FAILED – indicates that the command failed to execute successfully and occurs only if a driver continues to process commands after a command fails

in 2.1.0 version, driver will use a new protocol with server bulk protocol for batch insert when possible. (batch without Statement.RETURN_GENERATED_KEYS and streams) to have much faster batch when server MariaDB >= 10.2.7. According to environment, this can be 10 times faster.
When used, all inserts / updates are handled by bulk, driver then doesn't know individual row change, and return array of SUCCESS_NO_INFO ( =-2).

Hibernate do that sort of check too : https://github.com/hibernate/hibernate-orm/blob/master/hibernate-core/src/main/java/org/hibernate/jdbc/Expectations.java#L58 but check -2 (SUCCESS_NO_INFO) and -3 (EXECUTE_FAILED) values.

This new "bulk" functionality can be disabled setting option "useBulkStmts" to false.
And issue has to be created on gerrit/gwtorm

Could the new bulk be disabled by default instead please?

This is an efficient way to handle batch that correspond to standard.
The best is to correct error in gerrit.

I've submit a gerrit issue : https://bugs.chromium.org/p/gerrit/issues/detail?id=7105
During this time, behaviour can be corrected setting option "useBulkStmts" to false.