Ingest logs and data from a GCP Pub/Sub

Collect GCP Pub/Sub data with Cortex Cloud.

If you use the Pub/Sub messaging service from Global Cloud Platform (GCP), you can send logs and data from your GCP instance to Cortex Cloud. Data from GCP is then searchable in Cortex Cloud to provide additional information and context to your investigations using the GCP Cortex Query Language (XQL) dataset, which is dependent on the type of GCP logs collected. For example queries, refer to the in-app XQL Library. You can configure a Google Cloud Platform collector to receive generic, flow, audit, or Google Cloud DNS logs. When configuring generic logs, you can receive logs in a Raw, JSON, CEF, LEEF, Cisco, or Corelight format.

You can also configure Cortex Cloud to normalize different GCP logs as part of the enhanced cloud protection, which you can query with XQL Search using the applicable dataset. Cortex Cloud can also generate Cortex Cloud issues (Analytics, IOC, BIOC, and Correlation Rules), when relevant, from GCP logs. While Correlation Rules isssues are generated on non-normalized and normalized logs, Analytics, IOC, and BIOC issues are only raised on normalized logs.

Enhanced cloud protection provides the following:

• Normalization of cloud logs
• Cloud logs stitching
• Enrichment with cloud data
• Detection based on cloud analytics
• Cloud-tailored investigations

The following table lists the various GCP log types the XQL datasets you can use to query in XQL Search:

GCP log type	Dataset	Dataset with normalized data
Audit logs, including Google Kubernetes Engine (GKE) audit logs	google_cloud_logging_raw	cloud_audit_logs
Generic logs	Log Format types: CEF or LEEF: Automatically detected from either the logs or the user's input in the User Interface. Cisco: cisco_asa_raw Corelight: corelight_zeek_raw JSON or Raw: google_cloud_logging_raw	N/A
Google Cloud DNS logs	google_dns_raw	xdr_data: Once configured, Cortex Cloud ingests Google Cloud DNS logs as XDR network connection stories, which you can query with XQL Search using the xdr_data dataset with the preset called network_story.
Network flow logs	google_cloud_logging_raw	xdr_data: Once configured, Cortex Cloud ingests network flow logs as XDR network connection stories, which you can query with XQL Search using the xdr_data dataset with the preset called network_story.

Note

When collecting flow logs, we recommend that you include GKE annotations in your logs, which enable you to view the names of the containers that communicated with each other. GKE annotations are only included in logs if appended manually using the custom metadata configuration in GCP. For more information, see VPC Flow Logs Overview. In addition, to customize metadata fields, you must use the gcloud command-line interface or the API. For more information, see Using VPC Flow Logs.

To receive logs and data from GCP, you must first set up log forwarding using a Pub/Sub topic in GCP. You can configure GCP settings using either the GCP web interface or a GCP cloud shell terminal. After you set up your service account in GCP, you configure the Data Collection settings in Cortex Cloud. The setup process requires the subscription name and authentication key from your GCP instance.

After you set up log collection, Cortex Cloud immediately begins receiving new logs and data from GCP.

Set up log forwarding using the GCP web interface

• In Cortex Cloud, set up Data Collection.

  a. Navigate to Settings → Data Sources & Integrations.

  b. On the Data Sources & Integrations page, click + Add New, search for Google Cloud Platform, then hover over it and click Add.

  c. Specify the Subscription Name that you previously noted or copied.

  d. Browse to the JSON file containing your authentication key for the service account.

  e. Select the Log Type as one of the following, where your selection changes the options displayed.

  • Flow or Audit Logs: When selecting this log type, you can decide whether to normalize and enrich the logs as part of the enhanced cloud protection.

    • (Optional) You can Normalize and enrich flow and audit logs by selecting the checkbox (default). If selected, Cortex Cloud ingests the network flow logs as Cortex Cloud network connection stories, which you can query using XQL Search from the xdr_dataset dataset with the preset called network_story. In addition, you can configure Cortex Cloud to normalize GCP audit logs, which you can query with XQL Search using the cloud_audit_logs dataset.
    • The Vendor is automatically set to Google and Product to Cloud Logging, which is not configurable. This means that all GCP data for the flow and audit logs, whether it's normalized or not, can be queried in XQL Search using the google_cloud_logging_raw dataset.
  • Generic: When selecting this log type, you can configure the following settings.

    • Log Format: Select the log format type as Raw, JSON, CEF, LEEF, Cisco, or Corelight.

      • CEF or LEEF: The Vendor and Product defaults to Auto-Detect.

        Note

        For a Log Format set to CEF or LEEF, Cortex XDR reads events row by row to look for the Vendor and Product configured in the logs. When the values are populated in the event log row, Cortex XDR uses these values even if you specified a value in the Vendor and Product fields in the GCP data collector settings. Yet, when the values are blank in the event log row, Cortex XDR uses the Vendor and Product that you specified in the GCP data collector settings. If you did not specify a Vendor or Product in the GCP data collector settings, and the values are blank in the event log row, the values for both fields are set to unknown.
      • Cisco: The following fields are automatically set and not configurable.

        • Vendor: Cisco
        • Product: ASA

        Cisco data can be queried in XQL Search using the cisco_asa_raw dataset.
      • Corelight: The following fields are automatically set and not configurable.

        • Vendor: Corelight
        • Product: Zeek

        Corelight data can be queried in XQL Search using the corelight_zeek_raw dataset.
      • Raw or JSON: The following fields are automatically set and are configurable.

        • Vendor: Google
        • Product: Cloud Logging

        Raw or JSON data can be queried in XQL Search using the google_cloud_logging_raw dataset.Cortex XDR supports logs in single line format or multiline format. For a JSON format, multiline logs are collected automatically when the Log Format is configured as JSON. When configuring a Raw format, you must also define the Multiline Parsing Regex as explained below.
    • Vendor: (Optional) Specify a particular vendor name for the GCP generic data collection, which is used in the GCP XQL dataset <Vendor>_<Product>_raw that Cortex Cloud creates as soon as it begins receiving logs.
    • Product: (Optional) Specify a particular product name for the GCP generic data collection, which is used in the GCP XQL dataset name <Vendor>_<Product>_raw that Cortex Cloud creates as soon as it begins receiving logs.
    • Multiline Parsing Regex: (Optional) This option is only displayed when the Log Format is set to Raw, where you can set the regular expression that identifies when the multiline event starts in logs with multilines. It is assumed that when a new event begins, the previous one has ended.
  • Google Cloud DNS: When selecting this log type, you can configure whether to normalize the logs as part of the enhanced cloud protection.

    • Optional) You can Normalize DNS logs by selecting the checkbox (default). If selected, Cortex Cloud ingests the Google Cloud DNS logs as Cortex Cloud network connection stories, which you can query using XQL Search from the xdr_dataset dataset with the preset called network_story.
    • The Vendor is automatically set to Google and Product to DNS , which is not configurable. This means that all Google Cloud DNS logs, whether it's normalized or not, can be queried in XQL Search using the google_dns_raw dataset.

  f. Test the provided settings and, if successful, proceed to Enable log collection.

1. Log in to your GCP account.
2. Set up log forwarding from GCP to Cortex Cloud.

   a. Select Logging → Logs Router.

   b. Select Create Sink → Cloud Pub/Sub topic, and then click Next.

   c. To filter only specific types of data, select the filter or desired resource.

   d. In the Edit Sink configuration, define a descriptive Sink Name.

   e. Select Sink Destination → Create new Cloud Pub/Sub topic.

   f. Enter a descriptive Name that identifies the sink purpose for Cortex Cloud, and then Create.

   g. Create Sink and then Close when finished.
3. Create a subscription for your Pub/Sub topic.

   a. Select the menu icon in G Cloud, and then select Pub/Sub → Topics.

   b. Select the name of the topic you created in the previous steps. Use the filters if necessary.

   c. Select Create Subscription → Create subscription.

   d. Enter a unique Subscription ID.

   e. Choose Pull as the Delivery Type.

   f. Create the subscription. After the subscription is set up, G Cloud displays statistics and settings for the service.

   g. In the subscription details, identify and note your Subscription Name.

   Optionally, use the copy button to copy the name to the clipboard. You will need the name when you configure Collection in Cortex Cloud.
4. Create a service account and authentication key.

   You will use the key to enable Cortex Cloud to authenticate with the subscription service.

   a. Select the menu icon, and then select IAM & Admin → Service Accounts.

   b. Create Service Account.

   c. Enter a Service account name and then Create.

   d. Select a role for the account: Pub/Sub → Pub/Sub Subscriber.

   e. Click Continue → Done.

   f. Locate the service account by name, using the filters to refine the results, if needed.

   g. Click the Actions menu identified by the three dots in the row for the service account and then Create Key.

   h. Select JSON as the key type, and then Create. After you create the service account key, G Cloud automatically downloads it.
5. After Cortex Cloud begins receiving information from the GCP Pub/Sub service, you can use the XQL Query language to search for specific data.

Note

If you encounter errors while enabling this integration due to VPC Service Controls (VPC SC) in your Google Cloud organization, configure the required ingress rules. For detailed technical guidance, contact our Customer Support team.

Set up log forwarding using the GCP cloud shell terminal

In Cortex Cloud, set up Data Collection.

1. Navigate to Settings → Data Sources & Integrations.
2. On the Data Sources & Integrations page, click + Add New, search for Google Cloud Platform, then hover over it and click Add.
3. Specify the Subscription Name that you previously noted or copied.
4. Browse to the JSON file containing your authentication key for the service account.
5. Select the Log Type as one of the following, where your selection changes the options displayed.

   • Flow or Audit Logs: When selecting this log type, you can decide whether to normalize and enrich the logs as part of the enhanced cloud protection.

     • (Optional) You can Normalize and enrich flow and audit logs by selecting the checkbox (default). If selected, Cortex Cloud ingests the network flow logs as Cortex Cloud network connection stories, which you can query using XQL Search from the xdr_dataset dataset with the preset called network_story. In addition, you can configure Cortex Cloud to normalize GCP audit logs, which you can query with XQL Search using the cloud_audit_logs dataset.
     • The Vendor is automatically set to Google and Product to Cloud Logging, which is not configurable. This means that all GCP data for the flow and audit logs, whether it's normalized or not, can be queried in XQL Search using the google_cloud_logging_raw dataset.
   • Generic: When selecting this log type, you can configure the following settings.

     • Log Format: Select the log format type as Raw, JSON, CEF, LEEF, Cisco, or Corelight.

       • CEF or LEEF: The Vendor and Product defaults to Auto-Detect.

         For a Log Format set to CEF or LEEF, Cortex XDR reads events row by row to look for the Vendor and Product configured in the logs. When the values are populated in the event log row, Cortex XDR uses these values even if you specified a value in the Vendor and Product fields in the GCP data collector settings. Yet, when the values are blank in the event log row, Cortex XDR uses the Vendor and Product that you specified in the GCP data collector settings. If you did not specify a Vendor or Product in the GCP data collector settings, and the values are blank in the event log row, the values for both fields are set to unknown.
       • Cisco: The following fields are automatically set and not configurable.

         • Vendor: Cisco
         • Product: ASA

         Cisco data can be queried in XQL Search using the cisco_asa_raw dataset.
       • Corelight: The following fields are automatically set and not configurable.

         • Vendor: Corelight
         • Product: Zeek

         Corelight data can be queried in XQL Search using the corelight_zeek_raw dataset.
       • Raw or JSON: The following fields are automatically set and are configurable.

         • Vendor: Google
         • Product: Cloud Logging

         Raw or JSON data can be queried in XQL Search using the google_cloud_logging_raw dataset.

         Cortex Cloud supports logs in single line format or multiline format. For a JSON format, multiline logs are collected automatically when the Log Format is configured as JSON. When configuring a Raw format, you must also define the Multiline Parsing Regex as explained below.
     • Vendor: (Optional) Specify a particular vendor name for the GCP generic data collection, which is used in the GCP XQL dataset <Vendor>_<Product>_raw that Cortex Cloud creates as soon as it begins receiving logs.
     • Product: (Optional) Specify a particular product name for the GCP generic data collection, which is used in the GCP XQL dataset name <Vendor>_<Product>_raw that Cortex Cloud creates as soon as it begins receiving logs.
     • Multiline Parsing Regex: (Optional) This option is only displayed when the Log Format is set to Raw, where you can set the regular expression that identifies when the multiline event starts in logs with multilines. It is assumed that when a new event begins, the previous one has ended.
   • Google Cloud DNS: When selecting this log type, you can configure whether to normalize the logs as part of the enhanced cloud protection.

     • Optional) You can Normalize DNS logs by selecting the checkbox (default). If selected, Cortex Cloud ingests the Google Cloud DNS logs as Cortex Cloud network connection stories, which you can query using XQL Search from the xdr_dataset dataset with the preset called network_story.
     • The Vendor is automatically set to Google and Product to DNS , which is not configurable. This means that all Google Cloud DNS logs, whether it's normalized or not, can be queried in XQL Search using the google_dns_raw dataset.
6. Test the provided settings and, if successful, proceed to Enable log collection.
7. Launch the GCP cloud shell terminal or use your preferred shell with gcloud installed.

   

8. Define your project ID.

   gcloud config set project <PROJECT_ID>
                        

9. Create a Pub/Sub topic.

   gcloud pubsub topics create <TOPIC_NAME>
                        

10. Create a subscription for this topic.

    gcloud pubsub subscriptions create <SUBSCRIPTION_NAME> --topic=<TOPIC_NAME>
                         

    Note the subscription name you define in this step as you will need it to set up log ingestion from Cortex Cloud.
11. Create a logging sink.

    During the logging sink creation, you can also define additional log filters to exclude specific logs. To filter logs, supply the optional parameter --log-filter=<LOG_FILTER>

    gcloud logging sinks create <SINK_NAME> pubsub.googleapis.com/projects/<PROJECT_ID>/topics/<TOPIC_NAME> --log-filter=<LOG_FILTER>
                         

    If setup is successful, the console displays a summary of your log sink settings:

    Created [https://logging.googleapis.com/v2/projects/PROJECT_ID/sinks/SINK_NAME]. Please remember to grant `serviceAccount:LOGS_SINK_SERVICE_ACCOUNT` \ the Pub/Sub Publisher role on the topic. More information about sinks can be found at /logging/docs/export/configure_export

12. Grant log sink service account to publish to the new topic.

    Note the serviceAccount name from the previous step and use it to define the service for which you want to grant publish access.

    gcloud pubsub topics add-iam-policy-binding <TOPIC_NAME> --member serviceAccount:<LOGS_SINK_SERVICE_ACCOUNT> --role=roles/pubsub.publisher

13. Create a service account.

    For example, use cortex-xdr-sa as the service account name and Cortex Cloud Service Account as the display name.

    gcloud iam service-accounts create <SERVICE_ACCOUNT> --description="<DESCRIPTION>" --display-name="<DISPLAY_NAME>"

14. Grant the IAM role to the service account.

    gcloud pubsub subscriptions add-iam-policy-binding <SUBSCRIPTION_NAME> --member serviceAccount:<SERVICE_ACCOUNT>@<PROJECT_ID>.iam.gserviceaccount.com --role=roles/pubsub.subscriber

15. Create a JSON key for the service account.

    You will need the JSON file to enable Cortex Cloud to authenticate with the GCP service. Specify the file destination and filename using a .json extension.

    gcloud iam service-accounts keys create <OUTPUT_FILE> --iam-account <SERVICE_ACCOUNT>@<PROJECT_ID>.iam.gserviceaccount.com

16. After Cortex Cloud begins receiving information from the GCP Pub/Sub service, you can use the XQL Query language to search for specific data.