This documentation is for an out-of-date version of Apache Flink. We recommend you use the latest stable version.
SQL 客户端 #
Flink 的 Table & SQL API 可以处理 SQL 语言编写的查询语句,但是这些查询需要嵌入用 Java 或 Scala 编写的表程序中。此外,这些程序在提交到集群前需要用构建工具打包。这或多或少限制了 Java/Scala 程序员对 Flink 的使用。
SQL 客户端 的目的是提供一种简单的方式来编写、调试和提交表程序到 Flink 集群上,而无需写一行 Java 或 Scala 代码。SQL 客户端命令行界面(CLI) 能够在命令行中检索和可视化分布式应用中实时产生的结果。
入门 #
本节介绍如何在命令行里启动(setup)和运行你的第一个 Flink SQL 程序。
SQL 客户端捆绑在常规 Flink 发行版中,因此可以直接运行。它仅需要一个正在运行的 Flink 集群就可以在其中执行表程序。有关设置 Flink 群集的更多信息,请参见集群和部署部分。如果仅想试用 SQL 客户端,也可以使用以下命令启动本地集群:
./bin/start-cluster.sh
启动 SQL 客户端命令行界面 #
SQL Client 脚本也位于 Flink 的 bin 目录中。将来,用户可以通过启动嵌入式 standalone 进程或通过连接到远程 SQL 客户端网关来启动 SQL 客户端命令行界面。目前仅支持 embedded
,模式默认值embedded
。可以通过以下方式启动 CLI:
./bin/sql-client.sh
或者显式使用 embedded
模式:
./bin/sql-client.sh embedded
执行 SQL 查询 #
命令行界面启动后,你可以使用 HELP
命令列出所有可用的 SQL 语句。输入第一条 SQL 查询语句并按 Enter
键执行,可以验证你的设置及集群连接是否正确:
SELECT 'Hello World';
该查询不需要 table source,并且只产生一行结果。CLI 将从集群中检索结果并将其可视化。按 Q
键退出结果视图。
CLI 为维护和可视化结果提供三种模式。
表格模式(table mode)在内存中实体化结果,并将结果用规则的分页表格可视化展示出来。执行如下命令启用:
SET sql-client.execution.result-mode=table;
变更日志模式(changelog mode)不会实体化和可视化结果,而是由插入(+
)和撤销(-
)组成的持续查询产生结果流。
SET sql-client.execution.result-mode=changelog;
Tableau模式(tableau mode)更接近传统的数据库,会将执行的结果以制表的形式直接打在屏幕之上。具体显示的内容会取决于作业
执行模式的不同(execution.type
):
SET sql-client.execution.result-mode=tableau;
注意当你使用这个模式运行一个流式查询的时候,Flink 会将结果持续的打印在当前的屏幕之上。如果这个流式查询的输入是有限的数据集,
那么Flink在处理完所有的数据之后,会自动的停止作业,同时屏幕上的打印也会相应的停止。如果你想提前结束这个查询,那么可以直接使用
CTRL-C
按键,这个会停掉作业同时停止屏幕上的打印。
你可以用如下查询来查看三种结果模式的运行情况:
SELECT name, COUNT(*) AS cnt FROM (VALUES ('Bob'), ('Alice'), ('Greg'), ('Bob')) AS NameTable(name) GROUP BY name;
此查询执行一个有限字数示例:
变更日志模式 下,看到的结果应该类似于:
+ Bob, 1
+ Alice, 1
+ Greg, 1
- Bob, 1
+ Bob, 2
表格模式 下,可视化结果表将不断更新,直到表程序以如下内容结束:
Bob, 2
Alice, 1
Greg, 1
Tableau模式 下,如果这个查询以流的方式执行,那么将显示以下内容:
+-----+----------------------+----------------------+
| +/- | name | cnt |
+-----+----------------------+----------------------+
| + | Bob | 1 |
| + | Alice | 1 |
| + | Greg | 1 |
| - | Bob | 1 |
| + | Bob | 2 |
+-----+----------------------+----------------------+
Received a total of 5 rows
如果这个查询以批的方式执行,显示的内容如下:
+-------+-----+
| name | cnt |
+-------+-----+
| Alice | 1 |
| Bob | 2 |
| Greg | 1 |
+-------+-----+
3 rows in set
这几种结果模式在 SQL 查询的原型设计过程中都非常有用。这些模式的结果都存储在 SQL 客户端 的 Java 堆内存中。为了保持 CLI
界面及时响应,变更日志模式仅显示最近的 1000 个更改。表格模式支持浏览更大的结果,这些结果仅受可用主内存和配置的最大行数(sql-client.execution.max-table-result.rows
)的限制。
注意 在批处理环境下执行的查询只能用表格模式或者Tableau模式进行检索。
定义查询语句后,可以将其作为长时间运行的独立 Flink 作业提交给集群。配置部分解释如何声明读取数据的 table source,写入数据的 sink 以及配置其他表程序属性的方法。
Configuration #
SQL Client startup options #
The SQL Client can be started with the following optional CLI commands. They are discussed in detail in the subsequent paragraphs.
./bin/sql-client.sh --help
Mode "embedded" (default) submits Flink jobs from the local machine.
Syntax: [embedded] [OPTIONS]
"embedded" mode options:
-d,--defaults <environment file> Deprecated feature: the environment
properties with which every new
session is initialized. Properties
might be overwritten by session
properties.
-e,--environment <environment file> Deprecated feature: the environment
properties to be imported into the
session. It might overwrite default
environment properties.
-f,--file <script file> Script file that should be executed.
In this mode, the client will not
open an interactive terminal.
-h,--help Show the help message with
descriptions of all options.
-hist,--history <History file path> The file which you want to save the
command history into. If not
specified, we will auto-generate one
under your user's home directory.
-i,--init <initialization file> Script file that used to init the
session context. If get error in
execution, the sql client will exit.
Notice it's not allowed to add query
or insert into the init file.
-j,--jar <JAR file> A JAR file to be imported into the
session. The file might contain
user-defined classes needed for the
execution of statements such as
functions, table sources, or sinks.
Can be used multiple times.
-l,--library <JAR directory> A JAR file directory with which every
new session is initialized. The files
might contain user-defined classes
needed for the execution of
statements such as functions, table
sources, or sinks. Can be used
multiple times.
-pyarch,--pyArchives <arg> Add python archive files for job. The
archive files will be extracted to
the working directory of python UDF
worker. Currently only zip-format is
supported. For each archive file, a
target directory be specified. If the
target directory name is specified,
the archive file will be extracted to
a directory with the
specified name. Otherwise, the
archive file will be extracted to a
directory with the same name of the
archive file. The files uploaded via
this option are accessible via
relative path. '#' could be used as
the separator of the archive file
path and the target directory name.
Comma (',') could be used as the
separator to specify multiple archive
files. This option can be used to
upload the virtual environment, the
data files used in Python UDF (e.g.:
--pyArchives
file:///tmp/py37.zip,file:///tmp/data
.zip#data --pyExecutable
py37.zip/py37/bin/python). The data
files could be accessed in Python
UDF, e.g.: f = open('data/data.txt',
'r').
-pyexec,--pyExecutable <arg> Specify the path of the python
interpreter used to execute the
python UDF worker (e.g.:
--pyExecutable
/usr/local/bin/python3). The python
UDF worker depends on Python 3.6+,
Apache Beam (version == 2.27.0), Pip
(version >= 7.1.0) and SetupTools
(version >= 37.0.0). Please ensure
that the specified environment meets
the above requirements.
-pyfs,--pyFiles <pythonFiles> Attach custom files for job.
The standard resource file suffixes
such as .py/.egg/.zip/.whl or
directory are all supported. These
files will be added to the PYTHONPATH
of both the local client and the
remote python UDF worker. Files
suffixed with .zip will be extracted
and added to PYTHONPATH. Comma (',')
could be used as the separator to
specify multiple files (e.g.:
--pyFiles
file:///tmp/myresource.zip,hdfs:///$n
amenode_address/myresource2.zip).
-pyreq,--pyRequirements <arg> Specify a requirements.txt file which
defines the third-party dependencies.
These dependencies will be installed
and added to the PYTHONPATH of the
python UDF worker. A directory which
contains the installation packages of
these dependencies could be specified
optionally. Use '#' as the separator
if the optional parameter exists
(e.g.: --pyRequirements
file:///tmp/requirements.txt#file:///
tmp/cached_dir).
-s,--session <session identifier> The identifier for a session.
'default' is the default identifier.
-u,--update <SQL update statement> Deprecated Experimental (for testing
only!) feature: Instructs the SQL
Client to immediately execute the
given update statement after starting
up. The process is shut down after
the statement has been submitted to
the cluster and returns an
appropriate return code. Currently,
this feature is only supported for
INSERT INTO statements that declare
the target sink table.Please use
option -f to submit update statement.
SQL Client Configuration #
Key | Default | Type | Description |
---|---|---|---|
sql-client.execution.max-table-result.rowsBatch Streaming |
1000000 | Integer | The number of rows to cache when in the table mode. If the number of rows exceeds the specified value, it retries the row in the FIFO style. |
sql-client.execution.result-modeBatch Streaming |
TABLE | Enum Possible values: [TABLE, CHANGELOG, TABLEAU] |
Determine the mode when display the query result. The available values are ['table', 'tableau', 'changelog']. The 'table' mode materializes results in memory and visualizes them in a regular, paginated table representation. The 'changelog' mode does not materialize results and visualizes the result stream that is produced by a continuous query. The 'tableau' mode is more like a traditional way which will display the results in the screen directly with a tableau format. |
sql-client.verboseBatch Streaming |
false | Boolean | Determine whether to output the verbose output to the console. If set the option true, it will print the exception stack. Otherwise, it only output the cause. |
Initialize Session Using SQL Files #
A SQL query needs a configuration environment in which it is executed. SQL Client supports the -i
startup option to execute an initialization SQL file to setup environment when starting up the SQL Client.
The so-called initialization SQL file can use DDLs to define available catalogs, table sources and sinks,
user-defined functions, and other properties required for execution and deployment.
An example of such a file is presented below.
-- Define available catalogs
CREATE CATALOG MyCatalog
WITH (
'type' = 'hive'
);
USE CATALOG MyCatalog;
-- Define available database
CREATE DATABASE MyDatabase;
USE MyDatabase;
-- Define TABLE
CREATE TABLE MyTable(
MyField1 INT,
MyField2 STRING
) WITH (
'connector' = 'filesystem',
'path' = '/path/to/something',
'format' = 'csv'
);
-- Define VIEW
CREATE VIEW MyCustomView AS SELECT MyField2 FROM MyTable;
-- Define user-defined functions here.
CREATE FUNCTION foo.bar.AggregateUDF AS myUDF;
-- Properties that change the fundamental execution behavior of a table program.
SET table.planner = blink; -- planner: either 'blink' (default) or 'old'
SET execution.runtime-mode = streaming; -- execution mode either 'batch' or 'streaming'
SET sql-client.execution.result-mode = table; -- available values: 'table', 'changelog' and 'tableau'
SET sql-client.execution.max-table-result.rows = 10000; -- optional: maximum number of maintained rows
SET parallelism.default = 1; -- optional: Flink's parallelism (1 by default)
SET pipeline.auto-watermark-interval = 200; --optional: interval for periodic watermarks
SET pipeline.max-parallelism = 10; -- optional: Flink's maximum parallelism
SET table.exec.state.ttl=1000; -- optional: table program's idle state time
SET restart-strategy = fixed-delay;
-- Configuration options for adjusting and tuning table programs.
SET table.optimizer.join-reorder-enabled = true;
SET table.exec.spill-compression.enabled = true;
SET table.exec.spill-compression.block-size = 128kb;
This configuration:
- connects to Hive catalogs and uses
MyCatalog
as the current catalog withMyDatabase
as the current database of the catalog, - defines a table
MyTableSource
that can read data from a CSV file, - defines a view
MyCustomView
that declares a virtual table using a SQL query, - defines a user-defined function
myUDF
that can be instantiated using the class name, - uses the blink planner in streaming mode for running statements and a parallelism of 1,
- runs exploratory queries in the
table
result mode, - and makes some planner adjustments around join reordering and spilling via configuration options.
When using -i <init.sql>
option to initialize SQL Client session, the following statements are allowed in an initialization SQL file:
- DDL(CREATE/DROP/ALTER),
- USE CATALOG/DATABASE,
- LOAD/UNLOAD MODULE,
- SET command,
- RESET command.
When execute queries or insert statements, please enter the interactive mode or use the -f option to submit the SQL statements.
Attention If SQL Client meets errors in initialization, SQL Client will exit with error messages.
Dependencies #
The SQL Client does not require to setup a Java project using Maven or SBT. Instead, you can pass the
dependencies as regular JAR files that get submitted to the cluster. You can either specify each JAR
file separately (using --jar
) or define entire library directories (using --library
). For
connectors to external systems (such as Apache Kafka) and corresponding data formats (such as JSON),
Flink provides ready-to-use JAR bundles. These JAR files can be downloaded for each release from
the Maven central repository.
The full list of offered SQL JARs and documentation about how to use them can be found on the connection to external systems page.
Use SQL Client to submit job #
SQL Client allows users to submit jobs either within the interactive command line or using -f
option to execute sql file.
In both modes, SQL Client supports to parse and execute all types of the Flink supported SQL statements.
Interactive Command Line #
In interactive Command Line, the SQL Client reads user inputs and executes the statement when getting semicolon (;
).
SQL Client will print success message if the statement is executed successfully. When getting errors, SQL Client will also print error messages.
By default, the error message only contains the error cause. In order to print the full exception stack for debugging, please set the
sql-client.verbose
to true through command SET sql-client.verbose = true;
.
Execute SQL Files #
SQL Client supports to execute a SQL script file with the -f
option. SQL Client will execute
statements one by one in the SQL script file and print execution messages for each executed statements.
Once a statement is failed, the SQL Client will exist and all the remaining statements will not be executed.
An example of such a file is presented below.
CREATE TEMPORARY TABLE users (
user_id BIGINT,
user_name STRING,
user_level STRING,
region STRING,
PRIMARY KEY (user_id) NOT ENFORCED
) WITH (
'connector' = 'upsert-kafka',
'topic' = 'users',
'properties.bootstrap.servers' = '...',
'key.format' = 'csv',
'value.format' = 'avro'
);
-- set sync mode
SET table.dml-sync=true;
-- set the job name
SET pipeline.name=SqlJob;
-- set the queue that the job submit to
SET yarn.application.queue=root;
-- set the job parallelism
SET parallelism.default=100;
-- restore from the specific savepoint path
SET execution.savepoint.path=/tmp/flink-savepoints/savepoint-cca7bc-bb1e257f0dab;
INSERT INTO pageviews_enriched
SELECT *
FROM pageviews AS p
LEFT JOIN users FOR SYSTEM_TIME AS OF p.proctime AS u
ON p.user_id = u.user_id;
This configuration:
- defines a temporal table source
users
that reads from a CSV file, - set the properties, e.g job name,
- set the savepoint path,
- submit a sql job that load the savepoint from the specified savepoint path.
Attention Comparing to interactive mode, SQL Client will stop execution and exits when meets errors.
Execute a set of SQL statements #
SQL Client execute each INSERT INTO statement as a single Flink job. However, this is sometimes not
optimal because some part of the pipeline can be reused. SQL Client supports STATEMENT SET syntax to
execute a set of SQL statements. This is an equivalent feature with StatementSet in Table API. The
STATEMENT SET
syntax encloses one or more INSERT INTO
statements. All statements in a STATEMENT SET
block are holistically optimized and executed as a single Flink job. Joint optimization and execution
allows for reusing common intermediate results and can therefore significantly improve the efficiency
of executing multiple queries.
Syntax #
BEGIN STATEMENT SET;
-- one or more INSERT INTO statements
{ INSERT INTO|OVERWRITE <select_statement>; }+
END;
Attention The statements of enclosed in the STATEMENT SET
must be separated by a semicolon (;).
Flink SQL> CREATE TABLE pageviews (
> user_id BIGINT,
> page_id BIGINT,
> viewtime TIMESTAMP,
> proctime AS PROCTIME()
> ) WITH (
> 'connector' = 'kafka',
> 'topic' = 'pageviews',
> 'properties.bootstrap.servers' = '...',
> 'format' = 'avro'
> );
[INFO] Execute statement succeed.
Flink SQL> CREATE TABLE pageview (
> page_id BIGINT,
> cnt BIGINT
> ) WITH (
> 'connector' = 'jdbc',
> 'url' = 'jdbc:mysql://localhost:3306/mydatabase',
> 'table-name' = 'pageview'
> );
[INFO] Execute statement succeed.
Flink SQL> CREATE TABLE uniqueview (
> page_id BIGINT,
> cnt BIGINT
> ) WITH (
> 'connector' = 'jdbc',
> 'url' = 'jdbc:mysql://localhost:3306/mydatabase',
> 'table-name' = 'uniqueview'
> );
[INFO] Execute statement succeed.
Flink SQL> BEGIN STATEMENT SET;
[INFO] Begin a statement set.
Flink SQL> INSERT INTO pageviews
> SELECT page_id, count(1)
> FROM pageviews
> GROUP BY page_id;
[INFO] Add SQL update statement to the statement set.
Flink SQL> INSERT INTO uniqueview
> SELECT page_id, count(distinct user_id)
> FROM pageviews
> GROUP BY page_id;
[INFO] Add SQL update statement to the statement set.
Flink SQL> END;
[INFO] Submitting SQL update statement to the cluster...
[INFO] SQL update statement has been successfully submitted to the cluster:
Job ID: 6b1af540c0c0bb3fcfcad50ac037c862
CREATE TABLE pageviews (
user_id BIGINT,
page_id BIGINT,
viewtime TIMESTAMP,
proctime AS PROCTIME()
) WITH (
'connector' = 'kafka',
'topic' = 'pageviews',
'properties.bootstrap.servers' = '...',
'format' = 'avro'
);
CREATE TABLE pageview (
page_id BIGINT,
cnt BIGINT
) WITH (
'connector' = 'jdbc',
'url' = 'jdbc:mysql://localhost:3306/mydatabase',
'table-name' = 'pageview'
);
CREATE TABLE uniqueview (
page_id BIGINT,
cnt BIGINT
) WITH (
'connector' = 'jdbc',
'url' = 'jdbc:mysql://localhost:3306/mydatabase',
'table-name' = 'uniqueview'
);
BEGIN STATEMENT SET;
INSERT INTO pageviews
SELECT page_id, count(1)
FROM pageviews
GROUP BY page_id;
INSERT INTO uniqueview
SELECT page_id, count(distinct user_id)
FROM pageviews
GROUP BY page_id;
END;
Execute DML statements sync/async #
By default, SQL Client executes DML statements asynchronously. That means, SQL Client will submit a job for the DML statement to a Flink cluster, and not wait for the job to finish. So SQL Client can submit multiple jobs at the same time. This is useful for streaming jobs, which are long-running in general.
SQL Client makes sure that a statement is successfully submitted to the cluster. Once the statement is submitted, the CLI will show information about the Flink job.
Flink SQL> INSERT INTO MyTableSink SELECT * FROM MyTableSource;
[INFO] Table update statement has been successfully submitted to the cluster:
Cluster ID: StandaloneClusterId
Job ID: 6f922fe5cba87406ff23ae4a7bb79044
Attention The SQL Client does not track the status of the running Flink job after submission. The CLI process can be shutdown after the submission without affecting the detached query. Flink’s restart strategy
takes care of the fault-tolerance. A query can be cancelled using Flink’s web interface, command-line, or REST API.
However, for batch users, it’s more common that the next DML statement requires to wait util the
previous DML statement finishes. In order to execute DML statements synchronously, you can set
table.dml-sync
option true in SQL Client.
Flink SQL> SET table.dml-sync = true;
[INFO] Session property has been set.
Flink SQL> INSERT INTO MyTableSink SELECT * FROM MyTableSource;
[INFO] Submitting SQL update statement to the cluster...
[INFO] Execute statement in sync mode. Please wait for the execution finish...
[INFO] Complete execution of the SQL update statement.
Attention If you want to terminate the job, just type CTRL-C
to cancel the execution.
Start a SQL Job from a savepoint #
Flink supports to start the job with specified savepoint. In SQL Client, it’s allowed to use SET
command to specify the path of the savepoint.
Flink SQL> SET execution.savepoint.path=/tmp/flink-savepoints/savepoint-cca7bc-bb1e257f0dab;
[INFO] Session property has been set.
-- all the following DML statements will be restroed from the specified savepoint path
Flink SQL> INSERT INTO ...
When the path to savepoint is specified, Flink will try to restore the state from the savepoint when executing all the following DML statements.
Because the specified savepoint path will affect all the following DML statements, you can use RESET
command to reset this config option, i.e. disable restoring from savepoint.
Flink SQL> RESET execution.savepoint.path;
[INFO] Session property has been reset.
For more details about creating and managing savepoints, please refer to Job Lifecycle Management.
Define a Custom Job Name #
SQL Client supports to define job name for queries and DML statements through SET
command.
Flink SQL> SET pipeline.name= 'kafka-to-hive' ;
[INFO] Session property has been set.
-- all the following DML statements will use the specified job name.
Flink SQL> INSERT INTO ...
Because the specified job name will affect all the following queries and DML statements, you can also use RESET
command to reset this configuration, i.e. use default job names.
Flink SQL> RESET pipeline.name;
[INFO] Session property has been reset.
If the option pipeline.name
is not specified, SQL Client will generate a default name for the submitted job, e.g. insert-into_<sink_table_name>
for INSERT INTO
statements.
Compatibility #
To be compatible with before, SQL Client still supports to initialize with environment YAML file and allows to SET
the key in YAML file.
When set the key defined in YAML file, the SQL Client will print the warning messages to inform.
Flink SQL> SET execution.type = batch;
[WARNING] The specified key 'execution.type' is deprecated. Please use 'execution.runtime-mode' instead.
[INFO] Session property has been set.
-- all the following DML statements will be restored from the specified savepoint path
Flink SQL> INSERT INTO ...
When using SET
command to print the properties, the SQL Client will also print all the properties.
To distinguish the deprecated key, the sql client use the ‘[DEPRECATED]’ as the identifier.
Flink SQL>SET;
execution.runtime-mode=batch
sql-client.execution.result-mode=table
table.planner=blink
[DEPRECATED] execution.planner=blink
[DEPRECATED] execution.result-mode=table
[DEPRECATED] execution.type=batch
If you want to see more information about environment files, please refer to previous docs version
局限与未来 #
当前的 SQL 客户端仅支持嵌入式模式。在将来,社区计划提供基于 REST 的 SQL 客户端网关(Gateway) 的功能,详见 FLIP-24 和 FLIP-91。