Google CloudSQL PostgreSQL Sink

Writes records to a CloudSQL PostgreSQL table. Each record will be written to a row in the table. For example, you periodically build a recommendation model for products on your online store. The model is stored in a GCS bucket and you want to export the contents of the bucket to a CloudSQL PostgreSQL table where it can be served to your users.

Column names would be auto detected from input schema.

Configuration

Property

Macro Enabled?

Version Introduced

Description

Property

Macro Enabled?

Version Introduced

Description

Use Connection

No

6.7.0/1.8.0

Optional. Whether to use an existing connection. If you use a connection, connection related properties do not appear in the plugin properties.

Connection

Yes

6.7.0/1.8.0

Optional. Name of the connection to use. Project and service account information will be provided by the connection. You can also use the macro function ${conn(connection_name)}

JDBC Driver Name

No

 

Required. Name of the JDBC driver to use.

Default is cloudsql-postgresql.

CloudSQL Instance Type

No

 

Optional. Whether the CloudSQL instance to connect to is private or public. 

Default is Public.

Connection Name

Yes
(6.10.0)

 

Required. The CloudSQL instance to connect to in the format <PROJECT_ID>:<REGION>:<INSTANCE_NAME>. Can be found in the instance overview page.

Port

Yes

6.9.0/1.10.5

Optional. Port that PostgreSQL is running on.

Default is 5432.

Username

Yes

 

Optional. User identity for connecting to the specified database.

Password

Yes

 

Optional. Password to use to connect to the specified database.

Connection Arguments

Yes

 

Optional. A list of arbitrary string key/value pairs as connection arguments. These arguments will be passed to the JDBC driver as connection arguments for JDBC drivers that may need additional configurations.

Reference Name

No

 

Required. Name used to uniquely identify this sink for lineage, annotating metadata, etc.

Database

Yes (6.9.0/1.10.5)

 

Required. CloudSQL PostgreSQL database name.

Table Name

Yes

 

Required. Name of the table to export to.

Transaction Isolation Level

Yes

 

Transaction isolation level for queries run by this sink.

Default is TRANSACTION_READ_COMMITTED.

Connection Timeout

Yes

 

The timeout value used for socket connect operations. If connecting to the server takes longer than this value, the connection is broken.The timeout is specified in seconds and a value of zero means that it is disabled.

Data Type Mapping

All PostgreSQL specific data types mapped to string and can have multiple input formats and one ‘canonical’ output form. Please, refer to PostgreSQL data types documentation to figure out proper formats.

PostgreSQL Data Type

CDAP Schema Data Type

PostgreSQL Data Type

CDAP Schema Data Type

bigint

long

bigserial

long

bit(n)

string
string with ‘0’ and ‘1’ chars exact n length

bit varying(n)

string
string with ‘0’ and ‘1’ chars max n length

boolean

boolean

bytea

bytes

character

string

character varying

string

double precision

double

integer

int

numeric(precision, scale)/decimal(precision, scale)

decimal

real

float

smallint

int

smallserial

int

serial

int

text

string

date

date

time [ (p) ] [ without time zone ]

time

time [ (p) ] with time zone

string

timestamp [ (p) ] [ without time zone ]

timestamp

timestamp [ (p) ] with time zone

timestamp
stored in UTC format in database

xml

string

tsquery

string

tsvector

string

uuid

string

box

string

cidr

string

circle

string

inet

string

interval

string

json

string

jsonb

string

line

string

lseg

string

macaddr

string

macaddr8

string

money

string

path

string

point

string

polygon

string

Examples

Connecting to a public CloudSQL PostgreSQL instance

You want to write output records to “users” table of CloudSQL PostgreSQL database named “prod”, as “postgres” user with “postgres” password. Get the latest version of the CloudSQL socket factory jar with driver and dependencies here), and then configure plugin with:

Property

Value

Property

Value

Reference Name

sink1

Driver Name

cloudsql-postsgresql

Database

prod

CloudSQL Instance Type

Public

Connection Name

[PROJECT_ID]:[REGION]:[INSTANCE_NAME]

Import Query

"select id, name, email, phone from users;"

Number of Splits

1

Username

postgresql

Password

postgresql

Connecting to a private CloudSQL PostgreSQL instance

If you want to connect to a private CloudSQL PostgreSQL instance, create a Compute Engine VM that runs the CloudSQL Proxy docker image using the following command:

# Set the environment variables export PROJECT=[project_id] export REGION=[vm-region] export ZONE=`gcloud compute zones list --filter="name=${REGION}" --limit 1 --uri --project=${PROJECT}| sed 's/.*\///'` export SUBNET=[vpc-subnet-name] export NAME=[gce-vm-name] export POSTGRESQL_CONN=[postgresql-instance-connection-name] # Create a Compute Engine VM gcloud beta compute --project=${PROJECT_ID} instances create ${INSTANCE_NAME} --zone=${ZONE} --machine-type=g1-small --subnet=${SUBNE} --no-address --metadata=startup-script="docker run -d -p 0.0.0.0:3306:3306 gcr.io/cloudsql-docker/gce-proxy:1.16 /cloud_sql_proxy -instances=${POSTGRESQL_CONNECTION_NAME}=tcp:0.0.0.0:3306" --maintenance-policy=MIGRATE --scopes=https://www.googleapis.com/auth/cloud-platform --image=cos-69-10895-385-0 --image-project=cos-cloud

Optionally, you can promote the internal IP address of the VM running the Proxy image to a static IP using:

# Get the VM internal IP export IP=`gcloud compute instances describe ${NAME} --zone ${ZONE} | grep "networkIP" | awk '{print $2}'` # Promote the VM internal IP to static IP gcloud compute addresses create postgresql-proxy --addresses ${IP} --region ${REGION} --subnet ${SUBNET} echo Proxy IP: ${IP}

Get the latest version of the CloudSQL socket factory jar with driver and dependencies from here, and then configure plugin with:

Property

Value

Property

Value

Reference Name

sink1

Driver Name

cloudsql-postsgresql

Database

prod

CloudSQL Instance Type

Private

Connection Name

The poxy IP returned from the echo Proxy IP: ${IP} command.

Import Query

"select id, name, email, phone from users;"

Number of Splits

1

Username

postgresql

Password

postgresql

 

Created in 2020 by Google Inc.