Home

Awesome

go-mysql-elasticsearch is a service syncing your MySQL data into Elasticsearch automatically.

It uses mysqldump to fetch the origin data at first, then syncs data incrementally with binlog.

Call for Committer/Maintainer

Sorry that I have no enough time to maintain this project wholly, if you like this project and want to help me improve it continuously, please contact me through email (siddontang@gmail.com).

Requirement: In the email, you should list somethings(including but not limited to below) to make me believe we can work together.

Your GitHub ID The contributions to go-mysql-elasticsearch before, including PRs or Issues. The reason why you can improve go-mysql-elasticsearch.

Install

How to use?

Notice

Source

In go-mysql-elasticsearch, you must decide which tables you want to sync into elasticsearch in the source config.

The format in config file is below:

[[source]]
schema = "test"
tables = ["t1", t2]

[[source]]
schema = "test_1"
tables = ["t3", t4]

schema is the database name, and tables includes the table need to be synced.

If you want to sync all table in database, you can use asterisk(*).

[[source]]
schema = "test"
tables = ["*"]

# When using an asterisk, it is not allowed to sync multiple tables
# tables = ["*", "table"]

Rule

By default, go-mysql-elasticsearch will use MySQL table name as the Elasticserach's index and type name, use MySQL table field name as the Elasticserach's field name.
e.g, if a table named blog, the default index and type in Elasticserach are both named blog, if the table field named title, the default field name is also named title.

Notice: go-mysql-elasticsearch will use the lower-case name for the ES index and type. E.g, if your table named BLOG, the ES index and type are both named blog.

Rule can let you change this name mapping. Rule format in config file is below:

[[rule]]
schema = "test"
table = "t1"
index = "t"
type = "t"
parent = "parent_id"
id = ["id"]

    [rule.field]
    mysql = "title"
    elastic = "my_title"

In the example above, we will use a new index and type both named "t" instead of default "t1", and use "my_title" instead of field name "title".

Rule field types

In order to map a mysql column on different elasticsearch types you can define the field type as follows:

[[rule]]
schema = "test"
table = "t1"
index = "t"
type = "t"

    [rule.field]
    // This will map column title to elastic search my_title
    title="my_title"

    // This will map column title to elastic search my_title and use array type
    title="my_title,list"

    // This will map column title to elastic search title and use array type
    title=",list"

    // If the created_time field type is "int", and you want to convert it to "date" type in es, you can do it as below
    created_time=",date"

Modifier "list" will translates a mysql string field like "a,b,c" on an elastic array type '{"a", "b", "c"}' this is specially useful if you need to use those fields on filtering on elasticsearch.

Wildcard table

go-mysql-elasticsearch only allows you determind which table to be synced, but sometimes, if you split a big table into multi sub tables, like 1024, table_0000, table_0001, ... table_1023, it is very hard to write rules for every table.

go-mysql-elasticserach supports using wildcard table, e.g:

[[source]]
schema = "test"
tables = ["test_river_[0-9]{4}"]

[[rule]]
schema = "test"
table = "test_river_[0-9]{4}"
index = "river"
type = "river"

"test_river_[0-9]{4}" is a wildcard table definition, which represents "test_river_0000" to "test_river_9999", at the same time, the table in the rule must be same as it.

At the above example, if you have 1024 sub tables, all tables will be synced into Elasticsearch with index "river" and type "river".

Parent-Child Relationship

One-to-many join ( parent-child relationship in Elasticsearch ) is supported. Simply specify the field name for parent property.

[[rule]]
schema = "test"
table = "t1"
index = "t"
type = "t"
parent = "parent_id"

Note: you should setup relationship with creating the mapping manually.

Filter fields

You can use filter to sync specified fields, like:

[[rule]]
schema = "test"
table = "tfilter"
index = "test"
type = "tfilter"

# Only sync following columns
filter = ["id", "name"]

In the above example, we will only sync MySQL table tfiler's columns id and name to Elasticsearch.

Ignore table without a primary key

When you sync table without a primary key, you can see below error message.

schema.table must have a PK for a column

You can ignore these tables in the configuration like:

# Ignore table without a primary key
skip_no_pk_table = true

Elasticsearch Pipeline

You can use Ingest Node Pipeline to pre-process documents before indexing, like JSON string decode, merge fileds and more.

[[rule]]
schema = "test"
table = "t1"
index = "t"
type = "_doc"

# pipeline id
pipeline = "my-pipeline-id"

Node: you should create pipeline manually and Elasticsearch >= 5.0.

Why not other rivers?

Although there are some other MySQL rivers for Elasticsearch, like elasticsearch-river-jdbc, elasticsearch-river-mysql, I still want to build a new one with Go, why?

Todo

Donate

If you like the project and want to buy me a cola, you can through:

PayPal微信
[

Feedback

go-mysql-elasticsearch is still in development, and we will try to use it in production later. Any feedback is very welcome.

Email: siddontang@gmail.com