PBnJ

Deep Dive into PBnJ (Power, Boot, and Network Jobs)

PBnJ (Power, Boot, and Network Jobs) is a service within the Tinkerbell stack responsible for managing the power and boot state of bare-metal servers. It provides interfaces to control the power status (e.g., power on, power off, reboot) and boot order of the hardware, typically through the Baseboard Management Controller (BMC).

PBnJ Architecture Overview

PBnJ integrates with the BMC interfaces on bare-metal servers, which are specialized microcontrollers embedded on the motherboard. The BMC allows administrators to manage and monitor the physical state of the machine remotely, independent of the operating system installed on it.

Key concepts around PBnJ:

  1. Power Management: PBnJ can turn servers on or off, reboot them, or force a power cycle.
  2. Boot Management: PBnJ can configure the boot order of a server, ensuring that it boots from the correct device (e.g., network, disk, USB) during the provisioning process.
  3. Network Management: While PBnJ’s name suggests network management, it primarily focuses on power and boot state, with network management handled by other Tinkerbell components.

Core Components of PBnJ

  • BMC Interface: PBnJ communicates with the BMC using industry-standard protocols like IPMI (Intelligent Platform Management Interface), Redfish, or vendor-specific APIs. These protocols allow PBnJ to manage the power and boot settings of the server remotely.
  • Power Commands: PBnJ issues commands to the BMC to control the power state of the server, such as turning it on, off, or rebooting it.
  • Boot Configuration: PBnJ can set the boot order or force a server to boot from a specific device, which is crucial during the initial stages of provisioning.

PBnJ Power and Boot Flow

Here’s a detailed look at how PBnJ works within the Tinkerbell provisioning process:

  1. Server Identification:
  • PBnJ identifies the server it needs to manage using hardware details like the MAC address or IP address of the BMC.
  • The hardware details are usually registered in Tinkerbell’s hardware inventory and linked to the specific BMC interface.
  1. Power Management:
  • PBnJ issues commands to the BMC to control the server’s power state. For example, it might power on a server to start the provisioning process or power it off after provisioning is complete.
  1. Boot Order Configuration:
  • PBnJ can configure the server to boot from a specific device, such as the network (PXE boot) or the local disk. This ensures that the server boots into the correct environment for provisioning.
  1. Execution and Monitoring:
  • PBnJ executes the power and boot commands and monitors the status to ensure the commands are executed successfully. If a command fails, PBnJ can retry the operation or log the error for further investigation.

Working Examples with PBnJ

Let’s explore a practical example of how PBnJ is used to manage the power and boot state of a server during the Tinkerbell provisioning process.

1. Setting Up PBnJ

PBnJ is typically deployed within a Kubernetes cluster as part of the Tinkerbell stack. Here’s an example Kubernetes manifest to deploy PBnJ:

PBnJ Deployment Manifest
apiVersion: apps/v1
kind: Deployment
metadata:
  name: pbnj
  namespace: tink-system
spec:
  replicas: 1
  selector:
    matchLabels:
      app: pbnj
  template:
    metadata:
      labels:
        app: pbnj
    spec:
      containers:
      - name: pbnj
        image: quay.io/tinkerbell/pbnj:latest
        ports:
        - containerPort: 8080
          name: http

This manifest deploys PBnJ in the tink-system namespace, exposing it on port 8080.

Service Definition

To expose PBnJ within the Kubernetes cluster:

apiVersion: v1
kind: Service
metadata:
  name: pbnj
  namespace: tink-system
spec:
  selector:
    app: pbnj
  ports:
  - protocol: TCP
    port: 8080
    targetPort: 8080

This service allows other Tinkerbell components or external clients to communicate with PBnJ.

2. Registering Hardware with BMC Information

To manage a server using PBnJ, you need to register the server’s BMC details in Tinkerbell. Here’s an example of how this might look in a hardware definition YAML:

id: "b3a12e4c-5f0e-11ec-b1c3-0242ac130002"
metadata:
  facility: "on-prem"
  instance:
    device_1: "52:54:00:4e:15:2b"
    manufacturer: "Dell"
    plan: "c3.large.x86"
    state: "ready"
bmc:
  ip: "192.168.1.100"
  username: "admin"
  password: "adminpass"
  type: "ipmi"
network:
  interfaces:
    - dhcp:
        arch: "x86_64"
        mac: "52:54:00:4e:15:2b"

In this configuration:

  • BMC Details: The bmc section includes the IP address, username, password, and type of BMC interface (e.g., IPMI, Redfish). PBnJ uses these details to communicate with the BMC.
  • Hardware ID: The unique ID of the server, which PBnJ uses to identify the correct hardware to manage.

3. Issuing Power Commands

Using PBnJ’s API or through the Tink CLI, you can issue power commands to control the state of the server. Here’s an example of how to power on a server:

Power On the Server
curl -X POST http://pbnj:8080/v1/machine/power \
-H "Content-Type: application/json" \
-d '{
  "id": "b3a12e4c-5f0e-11ec-b1c3-0242ac130002",
  "action": "power_on"
}'

This command instructs PBnJ to power on the server identified by b3a12e4c-5f0e-11ec-b1c3-0242ac130002.

Power Off the Server

To power off the server:

curl -X POST http://pbnj:8080/v1/machine/power \
-H "Content-Type: application/json" \
-d '{
  "id": "b3a12e4c-5f0e-11ec-b1c3-0242ac130002",
  "action": "power_off"
}'

This command powers off the server.

4. Configuring Boot Order

PBnJ can also configure the boot order of a server, ensuring it boots from the correct device. For example, you might want to configure the server to boot from the network for PXE boot:

Set Boot to PXE
curl -X POST http://pbnj:8080/v1/machine/boot \
-H "Content-Type: application/json" \
-d '{
  "id": "b3a12e4c-5f0e-11ec-b1c3-0242ac130002",
  "boot_device": "pxe"
}'

This command configures the server to boot from PXE.

Set Boot to Disk

To configure the server to boot from the local disk:

curl -X POST http://pbnj:8080/v1/machine/boot \
-H "Content-Type: application/json" \
-d '{
  "id": "b3a12e4c-5f0e-11ec-b1c3-0242ac130002",
  "boot_device": "disk"
}'

This command sets the boot device to the local disk, ensuring the server boots from its installed operating system after provisioning.

5. Monitoring and Troubleshooting PBnJ

You can monitor PBnJ’s logs to troubleshoot any issues related to power or boot management. For example, if a server fails to power on or boot correctly, PBnJ’s logs might provide insight into the problem:

kubectl logs -n tink-system deployment/pbnj

This command retrieves the logs for the PBnJ deployment, where you can see the interactions with the BMC and any errors that occurred.

Advanced PBnJ Use Cases

  1. Automated Recovery: PBnJ can be integrated into automated recovery workflows, where it automatically reboots a server if a failure is detected. This is useful in environments where high availability is critical, and downtime must be minimized.
  2. Scheduled Power Management: PBnJ can be used to schedule power operations, such as shutting down servers during off-peak hours or powering them on before business hours begin. This helps in reducing energy costs and optimizing resource usage.
  3. Integration with Monitoring Systems: PBnJ can be integrated with monitoring systems to trigger power or boot operations based on certain conditions. For example, if a monitoring system detects that a server is unresponsive, it could trigger a reboot via PBnJ.

4

. Secure BMC Access: PBnJ can be configured to interact with BMCs over secure channels, using encryption and authentication methods that align with organizational security policies. This is particularly important in environments where sensitive data is processed.

Conclusion

PBnJ is an essential component of the Tinkerbell stack, providing the ability to manage the power and boot state of bare-metal servers. By integrating with the BMC interfaces on servers, PBnJ allows for remote and automated control of hardware, enabling smooth and efficient provisioning processes. Whether you need to power on a server, configure its boot order, or perform a power cycle, PBnJ offers a robust and flexible solution for managing the physical state of your infrastructure.