Skip to content

Version 0.0.17

Summary

Version 0.0.17 introduces a sibling read-only Ceph plugin packaged as netbox-ceph 0.0.1 from this repository. It is the third integration direction tracked by the Proxbox project, after read-only Proxmox → NetBox reflection (netbox-proxbox) and opt-in NetBox → Proxmox intent (also in netbox-proxbox, shipped in 0.0.15). Ceph inventory is Proxmox-managed only in v1: no direct Ceph Dashboard / Prometheus / RGW / RBD / external-cluster integration, and no NetBox → Ceph write path.

Tracking issues:

What's new

netbox-ceph sibling plugin

  • New wheel netbox-ceph 0.0.1 declared in netbox_ceph/pyproject.toml, installed alongside netbox-proxbox (required_plugins = ["netbox_proxbox"]).
  • Nine Django models — CephPluginSettings (singleton), CephCluster, CephDaemon, CephOSD, CephPool, CephFilesystem, CephCrushRule, CephFlag, CephHealthCheck — with permissive payload JSON fields so the UI can render upstream data without lossy serialization.
  • Initial migration 0001_initial depends on netbox_proxbox + extras and namespaces every UniqueConstraint under netbox_ceph_*.
  • NetBoxModelViewSet per model with http_method_names = ("get", "head", "options") mounted at /api/plugins/ceph/; generic ObjectListView / ObjectView per model with register_model_view; filter-only NetBoxModelFilterSetForm per model with lazy ProxmoxEndpoint queryset binding; child tabs on CephCluster for daemons, OSDs, and pools; plugin home view.

Branch-aware CephSyncJob

  • netbox_ceph.jobs.CephSyncJob runs on NetBox's default RQ queue with a 7200s job_timeout, the same long-budget convention as ProxboxSyncJob.
  • netbox_ceph.services.http_client calls proxbox-api's /ceph/status and /ceph/sync/{status,daemons,osds,pools,filesystems,crush,flags,full} plain JSON GETs; reuses netbox_proxbox.services.backend_context.get_fastapi_request_context so no new authentication or endpoint-resolution path is introduced.
  • netbox_ceph.services.branch_lifecycle re-exports the netbox-proxbox branch helpers and adds branching_enabled_settings() reading CephPluginSettings.get_solo() with defaults prefix="ceph-sync" and on_conflict="fail". When branching is unavailable or disabled, the job runs against main.
  • Per-resource progress is persisted under job.data["ceph_sync"] as a {params, runtime_seconds, response: {stages: [...]}} structure mirroring the Proxbox sync job's proxbox_sync key.
  • On error, the branch is left open for inspection; on success with a branch, merge_branch is called.

Documentation

Compatibility

  • NetBox: 4.5.8 – 4.6.99 (unchanged from 0.0.15).
  • Paired backend: proxbox-api >= 0.0.11 (same floor as 0.0.15's HA surface). The /ceph/* routes ship in the backend release that introduced the HA REST shim; no further backend bump is required.
  • Companion Proxmox SDK: proxmox-sdk with proxmox_sdk/ceph/ facade.

Out of scope for v1

Deferred to future releases:

  • Direct Ceph Dashboard API integration.
  • Prometheus metric ingestion.
  • RGW / S3 bucket inventory.
  • RBD image inventory.
  • External non-Proxmox Ceph clusters.
  • Any NetBox → Ceph write operation (pool create, OSD in/out, daemon start/stop, flag mutation, CephFS destroy, etc.).

Verification

  • uv run ruff check . — clean.
  • uv run pytest tests/test_ceph_*.py -q — 25 passed; AST/source-contract pattern (no Django bootstrap), matching the repo convention.
  • Repo-wide collection: 794 tests (769 prior + 25 new), no collection errors.
  • Live proxbox-api /ceph/status route returns HTTP 401 (auth-gated, i.e. registered) from the dev endpoint at http://10.0.30.207:8000/ceph/status.