Skip to main content

Does a Cribl Stream hybrid Worker need a public IP, and how can you remove it safely?

  • Nov 19, 2026 8:28 PM UTC
  • 0 replies
  • 1 view

Jessica Bracken

Objective

Determine whether a Cribl Stream hybrid Worker requires a public IP, then replace an existing Worker without one through a gradual migration with validation checks. A public IP is not required under these conditions.

Environment

  • Cribl Stream hybrid deployment with a Cribl.Cloud Leader
  • Customer-managed Worker host in Azure or another customer-managed cloud environment
  • Push senders and destinations whose required network paths are known

Procedure

1. Determine whether a hybrid Worker needs a public IP

A hybrid Worker does not require a public IP when its push senders are reachable through private network paths and the Worker can initiate all required Cribl.Cloud connections outbound.

Cribl's official documentation describes the required connectivity:

  • Cribl.Cloud Enterprise explains that hybrid Workers can run in public or private cloud instances or on-premises, while the Leader resides in Cribl.Cloud. It requires outbound connectivity to the Cribl.Cloud Leader on port 4200, outbound port 443 access to the Leader and https://cdn.cribl.io, and connectivity to https://cdn.cribl.io/telemetry/.
  • Bootstrap Workers from Leader documents the bootstrap requirements, including ongoing outbound communication to the Leader on port 4200 and outbound port 443 access to https://cdn.cribl.io and the Cribl.Cloud Leader.

A public IP or another reachable ingress path may still be required when external systems push data directly to the Worker. Verify the source locations and every required destination before removing public access.

2. Replace an existing Worker without a public IP

Use this gradual replacement process for an existing production Worker. Do not remove the public IP in place before the replacement Worker is validated.

  1. Verify that all push senders are internal to your network.
  2. Verify that the existing Worker can reach the Cribl.Cloud Leader outbound.
  3. Work with your network team to provide an explicit outbound path for a VM without a public IP, such as a NAT Gateway, Standard Load Balancer with outbound rules, or firewall/NVA with a route.
  4. Have your cloud administrator create a replacement VM with the same OS, size, and disk as the existing Worker.
  5. Place the replacement VM in a subnet with the required outbound path.
  6. Do not attach a public IP to the replacement VM.
  7. Allow inbound Syslog from internal senders in the network security group.
  8. Test outbound connectivity from the replacement VM to the Cribl.Cloud Leader on port 4200 and port 443.
  9. Test outbound HTTPS connectivity from the replacement VM to https://cdn.cribl.io on port 443.
  10. In Cribl Stream, navigate to Workers.
  11. Select Add/Update Worker Node.
  12. Select Linux > Add.
  13. In the Add Linux Worker modal, select the existing hybrid Worker Group in the Group field.
  14. Verify the Leader or Outpost hostname/IP field.
  15. Copy the generated script.
  16. Run the generated script on the replacement VM.
  17. Add sudo before bash if the generated script requires elevated privileges.
  18. Verify that the replacement node appears in the Worker Group, is healthy, and has the same version and config bundle as the existing Worker.
  19. Have your network team point one or two Syslog senders to the replacement Worker's private IP.
  20. In Cribl Stream, navigate to Monitoring > Data > Sources/Destinations.
  21. Verify Syslog events in and Cribl HTTP events out on the replacement Worker.
  22. Navigate to Monitoring > Logs.
  23. Verify that no destination or connection errors appear.
  24. Leave the replacement Worker running for an agreed period, such as a few hours or a full business cycle.
  25. Have your network team repoint the remaining senders to the replacement Worker.
  26. Verify that the old Worker's input falls to zero and the replacement Worker carries the full load. Note: If any sender uses UDP Syslog, perform the switch during a quiet window because events sent during the switch cannot be retransmitted.
  27. Keep the old Worker running but idle for a short soak period to preserve a quick rollback path.
  28. Have your administrator stop Cribl on the old host by running /opt/cribl/bin/cribl stop.
  29. Verify that the old Worker shows as offline in the Leader.
  30. Verify that data volume and downstream delivery are unchanged.
  31. Remove the old node from the Leader's view.
  32. Release the old public IP and decommission the old VM.

The replacement Worker may also require outbound access to *.s3.<region>.amazonaws.com on port 443 for config bundles, and to the Cribl-managed Worker Group address on port 10200 by default, or the port configured on the Cribl HTTP destination, for data forwarding.