Skip to content

[Nexthop] Distro CLI: Add proxy device infrastructure for testing #982

Draft
raghav-nexthop wants to merge 7 commits intofacebook:mainfrom
nexthop-ai:distro_cli_3.3_device_update
Draft

[Nexthop] Distro CLI: Add proxy device infrastructure for testing #982
raghav-nexthop wants to merge 7 commits intofacebook:mainfrom
nexthop-ai:distro_cli_3.3_device_update

Conversation

@raghav-nexthop
Copy link
Contributor

Pre-submission checklist

  • I've ran the linters locally and fixed lint errors related to the files I modified in this PR. You can install the linters by running pip install -r requirements-dev.txt && pre-commit install
  • pre-commit run

Summary

This PR adds Docker-based proxy device infrastructure for end-to-end testing of the device update functionality. The following features are provided:

  • FBOSS services simulation for testing
  • Isolated test environment using Docker
  • Btrfs snapshot testing without real hardware

Test Plan

Ran the unit tests. Integration test will be added in the next PR.

$ python3 -m pytest fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater -v

fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_acquire_artifacts_and_services PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_acquire_artifacts_forwarding_stack PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_get_services_from_component_services PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_update_requires_device_ip PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_validate_component_not_in_manifest PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_validate_component_with_no_services PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_validate_component_without_download_or_execute PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_validate_non_updatable_component PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_validate_success_forwarding_stack PASSED
fboss-image/distro_cli/tests/device_test.py::TestDeviceUpdater::test_validate_success_platform_stack PASSED

================================================================ 10 passed in 0.04s ================================================================

The Distro CLI getip and ssh commands make a best-effort attempt to
determine an IP for the given switch based on its MAC address and then
either return that IP or directly ssh to it.

It does so by connecting to the Distro Infra container and checking on
the configured interface for pre-existing IP neighbor/ARP entries. If
no such entries are found it will attempt a subnet ping to refresh the
Linux kernel neighbor entries.

In most cases this is sufficient after the system has PXE booted and
therefore the neighbor cache is filled.

Testing is as easy as starting the Distro Infra container:
```
$ ./distro_infra.sh --intf eth1 --persist-dir data
```

Then running the fboss-image device getip command with the appropriate
MAC address:
```
$ ./fboss-image device dc:da:4d:fc:ad:2d getip
[0.00s] Getting IP for device dc:da:4d:fc:ad:2d
10.250.33.2
```
The fboss-image device image command configures the Distro
Infrastructure to PXE boot the given image file for the given device
on the next reboot.

This replaces the manual steps previously necessary with the MVP
Distro Infrastucture.

Testing:

```
$ ./fboss-image device dc:da:4d:fc:ad:2d image nh-fboss-broadcom-xgs.tar
[0.00s] Setting image for device dc:da:4d:fc:ad:2d: nh-fboss-broadcom-xgs.tar
[0.07s] Using persistent directory: /home/travisb/work/upstream/fboss/fboss-image/distro_infra/data
[0.07s] Extracting image tarball to /home/travisb/work/upstream/fboss/fboss-image/distro_infra/data/dc-da-4d-fc-ad-2d...
[1.96s] Image extracted successfully to /home/travisb/work/upstream/fboss/fboss-image/distro_infra/data/dc-da-4d-fc-ad-2d
[1.96s] Enabling PXE boot for MAC address: dc-da-4d-fc-ad-2d
[2.09s] Successfully configured device dc:da:4d:fc:ad:2d with image nh-fboss-broadcom-xgs.tar
[2.09s] Device is ready for PXE boot
```

Then watch the switch PXE boot on the next boot:
```
>>Checking Media Presence......
>>Media Present......
>>Start PXE over IPv6 on MAC: DC-DA-4D-FC-AD-2D. Press ESC key to abort PXE boot..
  Station IP address is FC00:33:0:0:0:FB05:5030:774C
  Server IP address is FC00:33:0:0:0:0:0:1000
  NBP filename is ipxev6.efi
  NBP filesize is 1158144 Bytes

>>Checking Media Presence......
>>Media Present......
 Downloading NBP file...

  NBP file downloaded successfully.
iPXE initialising devices...
```
@raghav-nexthop raghav-nexthop requested a review from a team as a code owner March 6, 2026 01:27
@meta-cla meta-cla bot added the CLA Signed label Mar 6, 2026
@raghav-nexthop raghav-nexthop marked this pull request as draft March 6, 2026 01:34
@raghav-nexthop raghav-nexthop force-pushed the distro_cli_3.3_device_update branch from e92d720 to 0353f56 Compare March 6, 2026 01:45
travisb-nexthop and others added 4 commits March 9, 2026 16:43
- Add DeviceUpdater class with component validation
- Add COMPONENT_SERVICES mapping for updatable components
- Add test manifest and unit tests for validation logic
- Stub implementations for artifact acquisition, package creation, and transfer

This is the first step in implementing device update functionality.
- Implement full update workflow in DeviceUpdater.update()
- Add artifact acquisition via build or download
- Add update package creation with update_service.sh script
- Add SSH-based transfer and execution on device
- Add integration with distro_infra for device IP resolution
- Add comprehensive unit tests for update workflow
- Add fboss_init.sh support for distro-base snapshot creation
@raghav-nexthop raghav-nexthop force-pushed the distro_cli_3.3_device_update branch from 0353f56 to 819e64f Compare March 11, 2026 17:41
@raghav-nexthop raghav-nexthop marked this pull request as ready for review March 11, 2026 17:41
@raghav-nexthop raghav-nexthop marked this pull request as draft March 11, 2026 18:24
@raghav-nexthop raghav-nexthop force-pushed the distro_cli_3.3_device_update branch from 819e64f to 8f882c7 Compare March 11, 2026 19:11
- Add Docker-based proxy device container simulating FBOSS services
- Add btrfs snapshot support for service root directories
- Add systemd service templates for FBOSS services
- Add device initialization and entrypoint scripts
- Update device_test.py with proxy device integration tests

This enables end-to-end testing of device update functionality.
@raghav-nexthop raghav-nexthop force-pushed the distro_cli_3.3_device_update branch from 8f882c7 to 44e87d6 Compare March 11, 2026 19:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants