postgresql_request plugin

Article
07/02/2023

The postgresql_request plugin sends a SQL query to an Azure PostgreSQL Server network endpoint and returns the first rowset in the results. The query may return more than one rowset, but only the first rowset is made available for the rest of the Kusto query.

The plugin is invoked with the evaluate operator.

Important

The postgresql_request plugin is disabled by default. To enable the plugin, run the .enable plugin postgresql_request command. To see which plugins are enabled, use .show plugin management commands.

Syntax

evaluate postgresql_request ( ConnectionString , SqlQuery [, SqlParameters] ) [: OutputSchema]

Learn more about syntax conventions.

Parameters

Name	Type	Required	Description
ConnectionString	`string`	✔️	The connection string that points at the PostgreSQL Server network endpoint. See authentication and how to specify the network endpoint.
SqlQuery	`string`	✔️	The query that is to be executed against the SQL endpoint. Must return one or more row sets. Only the first set is made available for the rest of the query.
SqlParameters	`dynamic`		A property bag object that holds key-value pairs to pass as parameters along with the query.
OutputSchema			The names and types for the expected columns of the `postgresql_request` plugin output. Syntax: `(` ColumnName `:` ColumnType [`,` ...] `)`

Note

Specifying the OutputSchema is highly recommended, as it allows the plugin to be used in scenarios that might otherwise not work without it, such as a cross-cluster query. The OutputSchema can also enable multiple query optimizations.
An error is raised if the run-time schema of the first row set returned by the SQL network endpoint doesn't match the OutputSchema schema.

Authentication and authorization

To authorize a PostgreSQL Server network endpoint, you must specify the authorization information in the connection string. The supported authorization method is via username and password.

Set callout policy

The plugin makes callouts to the PostgreSQL database. Make sure that the cluster's callout policy enables calls of type postgresql to the target PostgreSqlDbUri.

The following example shows how to define the callout policy for PostgreSQL databases. We recommend restricting the callout policy to specific endpoints (my_endpoint1, my_endpoint2).

[
  {
    "CalloutType": "postgresql",
    "CalloutUriRegex": "my_endpoint1\\.postgres\\.database\\.azure\\.com",
    "CanCall": true
  },
  {
    "CalloutType": "postgresql",
    "CalloutUriRegex": "my_endpoint2\\.postgres\\.database\\.azure\\.com",
    "CanCall": true
  }
]

The following example shows a .alter callout policy command for postgresql CalloutType:

.alter cluster policy callout @'[{"CalloutType": "postgresql", "CalloutUriRegex": "\\.postgresql\\.database\\.azure\\.com", "CanCall": true}]'

Username and password authentication

The postgresql_request plugin only supports username and password authentication to the PostgreSQL server endpoint and doesn't integrate with Microsoft Entra authentication.

The username and password are provided as part of the connections string using the following parameters:

User ID=...; Password=...;

Warning

Confidential or guarded information should be obfuscated from connection strings and queries so that they are omitted from any Kusto tracing. For more information, see obfuscated string literals.

Encryption and server validation

For security, SslMode is unconditionally set to Required when connecting to a PostgreSQL server network endpoint. As a result, the server must be configured with a valid SSL/TLS server certificate.

Specify the network endpoint

Specify the PostgreSQL network endpoint as part of the connection string.

Syntax:

Host = FQDN [Port = Port]

Where:

FQDN is the fully qualified domain name of the endpoint.
Port is the TCP port of the endpoint.

Examples

SQL query to Azure PostgreSQL DB

The following example sends a SQL query to an Azure PostgreSQL database. It retrieves all records from public."Table", and then processes the results.

Note

This example shouldn't be taken as a recommendation to filter or project data in this manner. SQL queries should be constructed to return the smallest dataset possible.

evaluate postgresql_request(
    'Host=contoso.postgres.database.azure.com; Port = 5432;'
    'Database=Fabrikam;'
    h'User Id=USERNAME;'
    h'Password=PASSWORD;',
    'select * from public."Table"') : (Id: int, Name: string)
| where Id > 0
| project Name

SQL query to an Azure PostgreSQL database with modifications

The following example sends a SQL query to an Azure PostgreSQL database retrieving all records from public."Table", while appending another datetime column, and then processes the results on the Azure Data Explorer side. It specifies a SQL parameter (@param0) to be used in the SQL query.

evaluate postgresql_request(
    'Server=contoso.postgres.database.azure.com; Port = 5432;'
    'Database=Fabrikam;'
    h'User Id=USERNAME;'
    h'Password=PASSWORD;',
    'select *, @param0 as dt from public."Table"',
    dynamic({'param0': datetime(2020-01-01 16:47:26.7423305)})) : (Id: int, Name: string, dt: datetime)
| where Id > 0
| project Name

SQL query to an Azure PostgreSQL database without a query-defined output schema

The following example sends a SQL query to an Azure PostgreSQL database without an output schema. This is not recommended unless the schema is unknown, as it may impact the performance of the query

evaluate postgresql_request(
    'Host=contoso.postgres.database.azure.com; Port = 5432;'
    'Database=Fabrikam;'
    h'User Id=USERNAME;'
    h'Password=PASSWORD;',
    'select * from public."Table"')
| where Id > 0
| project Name

This capability isn't supported in Azure Monitor.

Share via