Orphan resources¶
Orphan resources are things you are still paying for but no longer use: a volume left unattached after an instance was deleted, a floating IP nobody reserved anymore, a snapshot whose source volume is gone. OpenWatch surfaces them in one place so you can review and clean them up.
What OpenWatch flags as an orphan¶
The detector looks across every project and region in your organisation and reports the following cases:
| Resource | Condition | Why it is wasteful |
|---|---|---|
| Volume (Cinder) | Status available with no attachment |
A detached disk keeps billing for its full size |
| Floating IP (Neutron) | No associated port | A reserved public IP that routes to nothing |
| Instance (Nova) | Status SHUTOFF |
Stopped, but its disk and reservations may still cost |
| Snapshot (Cinder) | Source volume no longer exists | Storage kept for a volume that is gone |
| Backup (Cinder) | Source volume no longer exists | Same as above |
| Load balancer (Octavia) | Provisioning status ERROR |
A failed load balancer that never became usable |
Each row shows the resource name, a short detail line (size, type, VIP, etc.), the reason it was flagged, its creation date (age), and an estimated monthly cost. For volumes, snapshots and backups the estimate is the size multiplied by the storage price; for floating IPs it is the monthly IP rate. Stopped instances and errored load balancers are listed with a zero estimate because their residual cost depends on your setup, but they are still worth reviewing.
How detection runs¶
Orphan detection does not run live on every page load. A background job (refresh_orphans) scans your resources and stores the result, and the scheduled job runs automatically every few hours.
- The first time you open the page, OpenWatch triggers a scan and shows a "running" state. The page refreshes on its own when the result lands.
- You can force a fresh scan at any time with the refresh action on the page.
Why a short delay is normal
Scanning Nova, Cinder and Neutron across several projects and regions takes a few seconds. Running it in the background keeps the page fast and avoids hammering your cloud's APIs.
Reviewing and acting¶
For each orphan you have three options.
Acknowledge¶
If a resource is intentionally kept (a warm standby volume, a reserved IP for a future migration), acknowledge it. Acknowledged resources are hidden from the default view so your list stays focused on real waste. You can add a short reason, acknowledge several at once, and later un-acknowledge to bring an item back into the list (use the "show acknowledged" view to find them).
Delete¶
OpenWatch can delete an orphan directly through the OpenStack API from the page.
Deletion is permanent and acts on your cloud
Deleting an orphan here removes it from your actual cloud, not just from OpenWatch. There is no undo. Confirm the resource is truly unused (check the name, age and project) before deleting. If you are unsure, acknowledge it instead, or delete it from your provider console where you have the full context.
Export¶
You can export the current orphan list as CSV to share with your team or to track cleanup outside OpenWatch.
Availability¶
Orphan detection is a gated feature. Whether it is enabled depends on your plan; see Editions and plan limits. For inventory browsing without the waste analysis, use the resource explorer.
A practical cleanup routine¶
- Open the orphans page and let the scan complete.
- Sort by estimated monthly cost and start from the top.
- Acknowledge anything kept on purpose, with a reason.
- Delete or clean up the rest, largest first.
- Re-run monthly. Orphans accumulate quietly as environments change.