Data Lake Queries

Panther API search operations

Overview

The Panther API supports the following data lake operations:

• Listing your data lake databases, tables, and columns

• Executing a data lake (Data Explorer) query using SQL

• Executing a Search query

• Canceling any currently-running query

• Fetching the details of any previously executed query

• Listing all currently running or previously-executed queries with optional filters

You can invoke Panther's API by using your Console's API Playground, or the GraphQL-over-HTTP API. Learn more about these methods on Panther API.

See the sections below for GraphQL queries, mutations, and end-to-end workflow examples around core data lake query operations.

Queries managed via the API must be written in SQL; they cannot use PantherFlow.

Common Data Lake query operations

Below are some of the most common GraphQL Data Lake query operations in Panther. These examples demonstrate the documents you have to send using a GraphQL client (or curl) to make a call to Panther's GraphQL API.

Database Entities

Listing all database entities

# `AllDatabaseEntities` is a nickname for the operation
query AllDatabaseEntities {
  dataLakeDatabases {
     name
     description
     tables {
       name
       description
       columns {
         name
         description
         type
       }
     }
   }
 }

Fetching the entities of a particular database

Executing queries

Executing a data lake (Data Explorer) query

# `IssueDataLakeQuery` is a nickname for the operation
mutation IssueDataLakeQuery {
  executeDataLakeQuery(input: {
    sql: "select * from panther_logs.public.aws_alb limit 50"
  }) {
     id # the unique ID of the query
  }
}

Executing an Indicator Search query

Canceling a query

Fetching results for a data lake or Search query

When you execute a data lake or Search query, it can take a few seconds to a few minutes for results to come back. To confirm that the query has completed, you must check the status of the query (polling).

You can use the following query to check the query status, while also fetching its results if available:

Fetching the first page of results

# `QueryResults` is a nickname for the operation
query QueryResults {
  dataLakeQuery(id: "1234-1234-1234-1234") { # the unique ID of the query
    message
    status
    results {
      edges {
        node
      }
    }
  }
}

Fetching subsequent pages of results

The expected values of status and results depend on the query's status:

• If the query is still running:

  • status will have a value of running

  • results will have a value of null

• If the query has failed:

  • status will have a value of failed

  • results will have a value of null and the error message will be available in the message key

• If the query has completed

  • status will have a value of succeeded

  • results will be populated

All of the above (along with the possible values for status) , along with additional fields you are allowed to request. Learn about the different ways to explore the Panther API schema here.

Fetching metadata around a data lake or Search query

In the example above, we requested the results of a Panther query. It is also possible to request additional metadata around the query.

In the following example, we request these metadata along the first page of results:

# `QueryMetadata` is a nickname for the operation
query QueryMetadata {
  dataLakeQuery(id: "1234-1234-1234-1234") { # the unique ID of the query
    name
    isScheduled
    issuedBy {
      ... on User {
        email
      }
      ... on APIToken {
        name
      } 
    }
    sql
    message
    status
    startedAt
    completedAt
    results {
      edges {
        node
      }
    }
  }
}

Listing data lake and Search queries

Fetching the first page

# `ListDataLakeQueries` is a nickname for the operation
query ListDataLakeQueries {
  dataLakeQueries {
    name
    isScheduled
    issuedBy {
      ... on User {
        email
      }
      ... on APIToken {
        name
    } 
    }
    sql
    message
    status
    startedAt
    completedAt
    results { # we're only fetching the first page of results for each query
      edges {
        node
      }
    }
  }

Fetching subsequent pages

Fetching a filtered set

End-to-end examples

Below, we will build on the Common Operations examples to showcase an end-to-end flow.

Execute a data lake (Data Explorer) Query

NodeJS

// npm install graphql graphql-request

import { GraphQLClient, gql } from 'graphql-request';

const client = new GraphQLClient(
  'YOUR_PANTHER_API_URL', 
  { headers: { 'X-API-Key': 'YOUR_API_KEY' } 
});

// `IssueQuery` is a nickname for the query. You can fully omit it.
const issueQuery = gql`
  mutation IssueQuery($sql: String!) {
    executeDataLakeQuery(input: { sql: $sql }) {
      id
    }
  }
`;

// `GetQueryResults` is a nickname for the query. You can fully omit it.
const getQueryResults = gql`
  query GetQueryResults($id: ID!, $cursor: String) {
    dataLakeQuery(id: $id) {
      message
      status
      results(input: { cursor: $cursor }) {
        edges {
          node
        }
        pageInfo {
          endCursor
          hasNextPage
        }
      }
    }
  }
`;

(async () => {
  try {
    // an accumulator that holds all result nodes that we fetch
    let allResults = [];
    // a helper to know when to exit the loop
    let hasMore = true;
    // the pagination cursor
    let cursor = null;

    // issue a query
    const mutationData = await client.request(issueQuery, {
      sql: 'select * from panther_logs.public.aws_alb limit 5',
    });

    // Start polling the query until it returns results. From there,
    // keep fetching pages until there are no more left
    do {
      const queryData = await client.request(getQueryResults, {
        id: mutationData.executeDataLakeQuery.id,
        cursor,
      });

      // if it's still running, print a message and keep polling
      if (queryData.dataLakeQuery.status === 'running') {
        console.log(queryData.dataLakeQuery.message);
        continue;
      }

      // if it's not running & it's not completed, then it's
      // either cancelled or it has errored out. In this case,
      // throw an exception
      if (queryData.dataLakeQuery.status !== 'succeeded') {
        throw new Error(queryData.dataLakeQuery.message);
      }

      allResults = [...allResults, ...queryData.dataLakeQuery.results.edges.map(edge => edge.node)];

      hasMore = queryData.dataLakeQuery.results.pageInfo.hasNextPage;
      cursor = queryData.dataLakeQuery.results.pageInfo.endCursor;
    } while (hasMore);

    console.log(`Your query returned ${allResults.length} result(s)!`);
  } catch (err) {
    console.error(err.response);
  }
})();

Python

Execute a Search query

NodeJS

// npm install graphql graphql-request

import { GraphQLClient, gql } from 'graphql-request';

const client = new GraphQLClient(
  'YOUR_PANTHER_API_URL', 
  { headers: { 'X-API-Key': 'YOUR_API_KEY' } 
});

// `IssueQuery` is a nickname for the query. You can fully omit it.
const issueQuery = gql`
  mutation IssueQuery($input: ExecuteIndicatorSearchQueryInput!) {
    executeIndicatorSearchQuery(input: $input) {
      id
    }
  }
`;

// `GetQueryResults` is a nickname for the query. You can fully omit it.
const getQueryResults = gql`
  query GetQueryResults($id: ID!, $cursor: String) {
    dataLakeQuery(id: $id) {
      message
      status
      results(input: { cursor: $cursor }) {
        edges {
          node
        }
        pageInfo {
          endCursor
          hasNextPage
        }
      }
    }
  }
`;

(async () => {
  try {
    // an accumulator that holds all result nodes that we fetch
    let allResults = [];
    // a helper to know when to exit the loop
    let hasMore = true;
    // the pagination cursor
    let cursor = null;

    // issue a query
    const mutationData = await client.request(issueQuery, {
      input: {         
        indicators: ["226103014039"],
        startTime: "2022-03-29T00:00:00.001Z",
        endTime: "2022-03-30T00:00:00.001Z",
        indicatorName: "p_any_aws_account_ids"
      }
    });

    // Keep fetching pages until there are no more left
    do {
      const queryData = await client.request(getQueryResults, {
        id: mutationData.executeIndicatorSearchQuery.id,
        cursor,
      });

      // if it's still running, print a message and keep polling
      if (queryData.dataLakeQuery.status === 'running') {
        console.log(queryData.dataLakeQuery.message);
        continue;
      }

      // if it's not running & it's not completed, then it's
      // either cancelled or it has errored out. In this case,
      // throw an exception
      if (queryData.dataLakeQuery.status !== 'succeeded') {
        throw new Error(queryData.dataLakeQuery.message);
      }

      allResults = [...allResults, ...queryData.dataLakeQuery.results.edges.map(edge => edge.node)];

      hasMore = queryData.dataLakeQuery.results.pageInfo.hasNextPage;
      cursor = queryData.dataLakeQuery.results.pageInfo.endCursor;
    } while (hasMore);

    console.log(`Your query returned ${allResults.length} result(s)!`);
  } catch (err) {
    console.error(err.response);
  }
})();

Python