Skip to main content

MySQL

This is a change data capture (CDC) connector that captures change events from a MySQL database via the Binary Log.

It is available for use in the Flow web application. For local development or open-source workflows, ghcr.io/estuary/source-mysql:dev provides the latest version of the connector as a Docker image. You can also follow the link in your browser to see past image versions.

Supported platforms​

This connector supports MySQL on major cloud providers, as well as self-hosted instances.

Setup instructions are provided for the following platforms:

Prerequisites​

To use this connector, you'll need a MySQL database setup with the following.

  • The binlog_format system variable must be set to ROW (the default value).
  • The binary log expiration period should be at least 7 days.
    • This value may be set lower if necessary, but we discourage doing so as this may increase the likelihood of unrecoverable failures.
  • A database user with appropriate permissions:
    • REPLICATION CLIENT and REPLICATION SLAVE privileges.
    • Permission to read the tables being captured.
    • Permission to read from information_schema tables, if automatic discovery is used.
  • If the table(s) to be captured include columns of type DATETIME, the time_zone system variable must be set to an IANA zone name or numerical offset or the capture configured with a timezone to use by default.
Configuration Tip

To configure this connector to capture data from databases hosted on your internal network, you must set up SSH tunneling. For more specific instructions on setup, see configure connections with SSH tunneling.

Setup​

To meet these requirements, follow the steps for your hosting type.

Self-hosted MySQL​

  1. Create the flow_capture user with replication permission, and the ability to read all tables.

The SELECT permission can be restricted to just the tables that need to be captured, but automatic discovery requires information_schema access as well.

CREATE USER IF NOT EXISTS flow_capture
  IDENTIFIED BY 'secret'
  COMMENT 'User account for Flow MySQL data capture';
GRANT REPLICATION CLIENT, REPLICATION SLAVE ON *.* TO 'flow_capture';
GRANT SELECT ON *.* TO 'flow_capture';
  1. Configure the binary log to retain data for at least 7 days. We recommend 30 days where possible.
SET PERSIST binlog_expire_logs_seconds = 2592000;
  1. Configure the database's time zone. See below for more information.
SET PERSIST time_zone = '-05:00'

Amazon Aurora​

You must apply some of the settings to the entire Aurora DB cluster, and others to a database instance within the cluster. For each step, take note of which entity you're working with.

  1. Allow connections between the database and Estuary Flow. There are two ways to do this: by granting direct access to Flow's IP or by creating an SSH tunnel.

    1. To allow direct access:

      • Modify the instance, choosing Publicly accessible in the Connectivity settings.
      • Edit the VPC security group associated with your instance, or create a new VPC security group and associate it with the instance as described in the Amazon documentation. Create a new inbound rule and a new outbound rule that allow all traffic from the Estuary Flow IP addresses.

    2. To allow secure connections via SSH tunneling:

  2. Create a RDS parameter group to enable replication on your Aurora DB cluster.

    1. Create a parameter group. Create a unique name and description and set the following properties:

      • Family: aurora-mysql8.0
      • Type: DB ClusterParameter group

    2. Modify the new parameter group and update the following parameters:

      • binlog_format: ROW
      • binlog_row_metadata: FULL
      • read_only: 0

    3. Associate the parameter group with the DB cluster. While you're modifying the cluster, also set Backup Retention Period to 7 days.

    4. Reboot the cluster to allow the changes to take effect.

  3. Switch to your MySQL client. Run the following commands to create a new user for the capture with appropriate permissions:

CREATE USER IF NOT EXISTS flow_capture
  IDENTIFIED BY 'secret'
  COMMENT 'User account for Flow MySQL data capture';
GRANT REPLICATION CLIENT, REPLICATION SLAVE ON *.* TO 'flow_capture';
GRANT SELECT ON *.* TO 'flow_capture';
  1. Run the following command to set the binary log retention to 7 days, the maximum value Aurora permits:
CALL mysql.rds_set_configuration('binlog retention hours', 168);
  1. In the RDS console, note the instance's Endpoint and Port. You'll need these for the address property when you configure the connector.

Azure Database for MySQL​

  1. Allow connections between the database and Estuary Flow. There are two ways to do this: by granting direct access to Flow's IP or by creating an SSH tunnel.

    1. To allow direct access:

    2. To allow secure connections via SSH tunneling:

  2. Set the binlog_expire_logs_seconds server perameter to 2592000.

  3. Using MySQL workbench or your preferred client, create the flow_capture user with replication permission, and the ability to read all tables.

The SELECT permission can be restricted to just the tables that need to be captured, but automatic discovery requires information_schema access as well.

tip

Your username must be specified in the format username@servername.

CREATE USER IF NOT EXISTS flow_capture
  IDENTIFIED BY 'secret'
  COMMENT 'User account for Flow MySQL data capture';
GRANT REPLICATION CLIENT, REPLICATION SLAVE ON *.* TO 'flow_capture';
GRANT SELECT ON *.* TO 'flow_capture';
  1. Note the instance's host under Server name, and the port under Connection Strings (usually 3306). Together, you'll use the host:port as the address property when you configure the connector.

Capturing from Read Replicas​

This connector supports capturing from a read replica of your database, provided that binary logging is enabled on the replica and all other requirements are met.

Setting the MySQL time zone​

MySQL's time_zone server system variable is set to SYSTEM by default.

If you intend to capture tables including columns of the type DATETIME, and time_zone is set to SYSTEM, Flow won't be able to detect the time zone and convert the column to RFC3339 format. To avoid this, you must explicitly set the time zone for your database.

You can:

  • Specify a numerical offset from UTC.

    • For MySQL version 8.0.19 or higher, values from -13:59 to +14:00, inclusive, are permitted.
    • Prior to MySQL 8.0.19, values from -12:59 to +13:00, inclusive, are permitted

  • Specify a named timezone in IANA timezone format.

  • If you're using Amazon Aurora, create or modify the DB cluster parameter group associated with your MySQL database. Set the time_zone parameter to the correct value.

For example, if you're located in New Jersey, USA, you could set time_zone to -05:00 or -04:00, depending on the time of year. Because this region observes daylight savings time, you'd be responsible for changing the offset. Alternatively, you could set time_zone to America/New_York, and time changes would occur automatically.

If using IANA time zones, your database must include time zone tables. Learn more in the MySQL docs.

Capture Timezone Configuration

If you are unable to set the time_zone in the database and need to capture tables with DATETIME columns, the capture can be configured to assume a time zone using the timezone configuration property (see below). The timezone configuration property can be set as a numerical offset or IANA timezone format.

Backfills and performance considerations​

When the a MySQL capture is initiated, by default, the connector first backfills, or captures the targeted tables in their current state. It then transitions to capturing change events on an ongoing basis.

This is desirable in most cases, as in ensures that a complete view of your tables is captured into Flow. However, you may find it appropriate to skip the backfill, especially for extremely large tables.

In this case, you may turn of backfilling on a per-table basis. See properties for details.

Configuration​

You configure connectors either in the Flow web app, or by directly editing the catalog specification file.

See connectors to learn more about using connectors. The values and specification sample below provide configuration details specific to the MySQL source connector.

Properties​

Endpoint​

PropertyTitleDescriptionTypeRequired/Default
/addressServer AddressThe host or host:port at which the database can be reached.stringRequired
/userLogin UserThe database user to authenticate as.stringRequired, "flow_capture"
/passwordLogin PasswordPassword for the specified database user.stringRequired
/timezoneTimezoneTimezone to use when capturing datetime columns. Should normally be left blank to use the database's 'time_zone' system variable. Only required if the 'time_zone' system variable cannot be read and columns with type datetime are being captured. Must be a valid IANA time zone name or +HH:MM offset. Takes precedence over the 'time_zone' system variable if both are set.string
/advanced/dbnameDatabase NameThe name of database to connect to. In general this shouldn't matter. The connector can discover and capture from all databases it's authorized to access.string"mysql"
/advanced/node_idNode IDNode ID for the capture. Each node in a replication cluster must have a unique 32-bit ID. The specific value doesn't matter so long as it is unique. If unset or zero the connector will pick a value.integer
/advanced/skip_backfillsSkip BackfillsA comma-separated list of fully-qualified table names which should not be backfilled.string
/advanced/backfill_chunk_sizeBackfill Chunk SizeThe number of rows which should be fetched from the database in a single backfill query.integer131072
/advanced/skip_binlog_retention_checkSkip Binlog Retention Sanity CheckBypasses the 'dangerously short binlog retention' sanity check at startup. Only do this if you understand the danger and have a specific need.boolean

Bindings​

PropertyTitleDescriptionTypeRequired/Default
/namespaceNamespaceThe database/schema in which the table resides.stringRequired
/streamStreamName of the table to be captured from the database.stringRequired
info

When you configure this connector in the web application, the automatic discovery process sets up a binding for most tables it finds in your database, but there are exceptions.

Tables in the MySQL system schemas information_schema, mysql, performance_schema, and sys will not be discovered. You can add bindings for such tables manually.

Sample​

A minimal capture definition will look like the following:

captures:
  ${PREFIX}/${CAPTURE_NAME}:
    endpoint:
      connector:
        image: ghcr.io/estuary/source-mysql:dev
        config:
          address: "127.0.0.1:3306"
          user: "flow_capture"
          password: "secret"
    bindings:
      - resource:
          namespace: ${TABLE_NAMESPACE}
          stream: ${TABLE_NAME}
        target: ${PREFIX}/${COLLECTION_NAME}

Your capture definition will likely be more complex, with additional bindings for each table in the source database.

Learn more about capture definitions.

Troubleshooting Capture Errors​

The source-mysql connector is designed to halt immediately if something wrong or unexpected happens, instead of continuing on and potentially outputting incorrect data. What follows is a non-exhaustive list of some potential failure modes, and what action should be taken to fix these situations:

Unsupported Operations​

If your capture is failing with an "unsupported operation {ALTER,DROP,TRUNCATE,etc} TABLE" error, this indicates that such an operation has taken place impacting a table which is currently being captured.

In the case of DROP TABLE and other destructive operations this is not supported, and can only be resolved by removing the offending table(s) from the capture bindings list, after which you may recreate the capture if desired (causing the latest state of the table to be recaptured in its entirety).

In the case of ALTER TABLE we currently support table alterations to add or drop columns from a table. This error indicates that whatever alteration took place is not currently supported. Practically speaking the immediate resolution is the same as for a DROP or TRUNCATE TABLE, but if you frequently perform schema migrations it may be worth reaching out to see if we can add support for whatever table alteration you just did.

Data Manipulation Queries​

If your capture is failing with an "unsupported DML query" error, this means that an INSERT, UPDATE, DELETE or other data manipulation query is present in the MySQL binlog. This should generally not happen if binlog_format = 'ROW' as described in the Prerequisites section.

Resolving this error requires fixing the binlog_format system variable, and then either tearing down and recreating the entire capture so that it restarts at a later point in the binlog, or in the case of an INSERT/DELETE query it may suffice to remove the capture binding for the offending table and then re-add it.

Unhandled Queries​

If your capture is failing with an "unhandled query" error, some SQL query is present in the binlog which the connector does not (currently) understand.

In general, this error suggests that the connector should be modified to at least recognize this type of query, and most likely categorize it as either an unsupported DML Query, an unsupported Table Operation, or something that can safely be ignored. Until such a fix is made the capture cannot proceed, and you will need to backfill all collections to allow the capture to jump ahead to a later point in the binlog.

Metadata Errors​

If your capture is failing with a "metadata error" then something has gone badly wrong with the capture's tracking of table metadata, such as column names or datatypes.

This should never happen, and most likely means that the MySQL binlog itself is corrupt in some way. If this occurs, it can be resolved by backfilling all collections from the source.

Insufficient Binlog Retention​

If your capture fails with a "binlog retention period is too short" error, it is informing you that the MySQL binlog retention period is set to a dangerously low value.

The concern is that if a capture is disabled or the server becomes unreachable for longer than the binlog retention period, the database might delete a binlog segment which the capture isn't yet done with. If this happens then change events have been permanently lost, and the only way to get the capture running again is to skip ahead to a portion of the binlog which still exists. For correctness this requires backfilling the current contents of all tables from the source, and so we prefer to avoid it as much as possible. It's much easier to just set up your binlog retention with enough wiggle room to recover from temporary failures.

The "binlog retention period is too short" error should normally be fixed by setting a longer retention period as described in these setup instructions. However, advanced users who understand the risks can use the skip_binlog_retention_check configuration option to disable this safety.

Empty Collection Key​

Every Flow collection must declare a key which is used to group its documents. When testing your capture, if you encounter an error indicating collection key cannot be empty, you will need to either add a key to the table in your source, or manually edit the generated specification and specify keys for the collection before publishing to the catalog as documented here.