MySqlConnector | MySQL Connector for .NET | DevOps library

resume that connector once it is back?

After you run the connect-distributed script, it should rebalance the tasks.

Otherwise, there's a /restart API endpoint

if one worker goes down, it should Automatically assign the task to another worker

In my experience, when tasks actually fail, they remain failed, and the restart endpoint needs hit, if it's a temp failure and the logs don't show anything useful. However, your errors.tolerance setting may help isolate the problem somewhat

What you really need is a list of the missing codes in the user_type table. You can get that like this:

According to the MySqlConnector disscussion, it is not possible to return output direction parameters through it. You can try to write ADO.NET code to omit this restriction.

As @Kiran Joshi's comment I can use raw SQL to pause foreign key check and the MySQL version is:

migrationBuilder.Sql("SET FOREIGN_KEY_CHECKS=0");

Put this and migrationBuilder.Sql("SET FOREIGN_KEY_CHECKS=1"); (maybe not required) in Up and Down does the work.

When you go on mysql connection strings

You will see connection string with TCP port in it like this, try with this connection string construction:

Server=myServerAddress;Port=1234;Database=myDataBase;Uid=myUsername;Pwd=myPassword;

Also are you sure you are using right port?

The reason you are getting that exception since switching to a DbContext factory is because you are taking an entity and telling a completely new DbContext instance created by the factory to persist it, and any entities associated with it.

The issue you are seeing most certainly isn't with the entity you are trying to update, but child, or referenced entities that entity has. In your case if you are trying to persist a Character, and that Character has a collection of Personalities, then since that DBContext isn't tracking those instances of Personalities, it will treat them as if they are new entities and try and insert them.

Working with detached entities is frankly a pain. It might look like a simple matter of passing an entity to a new DbContext instance and calling Update, but aside from the simplest scenarios it is rarely that simple. Trying to build a data access layer as a Generic implementation just adds to the complexity.

At the most basic level, when processing an entity that has relations you need to associate those related entities to the DbContext when persisting. However, first you need to check that the DbContext instance in question isn't already tracking a matching entity and substitute the reference if it is.

The most common reason for this error is that the server timed out and closed the connection. This could happen for large datasets. Try using a chunksize:

For example, a column with datetime format in mysql is replicated as bigint

This is due to the default time.precision.mode used by the Debezium connector on the source side. If you look at the documentation, you'll notice that the default precision emits datetime columns as INT64, which explains why the sink connector writes the contents as a bigint.

You can set the time.precision.mode to connect on the source side for now so that the values can be properly interpreted by the JDBC sink connector.

foreign keys are not created

That's to be expected, see this Confluent GitHub Issue. At this time, the JDBC sink does not have the capabilities to support materializing Foreign Key relationships at the JDBC level.

order of columns is not preserved

That is also to be expected. There is no expected guarantee that Debezium should store the relational columns in the exact same order as they are in the database (although we do) and the JDBC sink connector is under no guarantee to retain the order of the fields as they're read from the emitted event. If the sink connector uses some container like a HashMap to store column names, it's plausible that the order would be very different than the source database.

If there is a necessity to retain a higher level of relational metadata such as foreign keys and column order at the destination system that mirrors that of the source system, you may need to look into a separate toolchain to replicate the initial schema and relationships through some type of schema dump, translation, and import to your destination database and then rely on the CDC pipeline for the data replication aspect.

This may be cause by a failure to negotiate TLS 1.3. Try setting TLS Version in your connection string: