Manage Storage and Usage
Use Admin → Storage Locations for your tenant's managed storage location, Admin → Storage for retained data and capacity, and Admin → Usage for accepted usage over the displayed time window.
Before you start
Sign in as a tenant administrator and confirm that you are working in the intended account. Custd assigns the storage prefix and routing. These screens do not require you to enter a bucket password, access key or other infrastructure credential.
Provision a storage location
- Open Admin → Storage Locations.
- Check the existing rows before making a change. If the page says No storage locations yet, choose Provision storage location.
- Wait for the provisioning request to finish. Inspect the returned Code, Assigned prefix, Provider, Region, and Status.
- Check Recent storage history for the corresponding change and actor. A history-loading error is separate from the location's status.
The screen provisions your tenant's isolated prefix in the environment's managed storage. For example, after provisioning, record the assigned code shown in the row; use that exact code when identifying the location in a support request. Do not invent a code or provide your own storage prefix.
A pending, failed or rejected status is not a successful validation. Reload after the request completes and check any error before relying on the location for new writes. Do not repeatedly provision locations to clear an error.
Inspect retained data and capacity
Open Admin → Storage and check the account name and Measured time. Read these values together:
| Value | Meaning |
|---|---|
| Data retained | The reported retained logical data size. |
| Events retained | The reported retained event count. |
| Limit | The effective storage allowance. |
| Remaining | The reported remaining allowance. |
| Measurement | The completeness state of the measurement. |
No storage usage measured means no snapshot is available yet; it does not prove zero usage. An old measurement or incomplete state needs a fresh, complete measurement before you treat it as a current capacity assessment.
Tenant administrators can inspect these values. Edit limit is a system administrator control; this tenant screen does not grant capacity-editing authority. If the Warning badge appears, review the measurement and your retention needs before sending more data.
What a hard limit actually does
A storage limit is an allowance, not just a number to watch. When a system administrator enables the hard limit for your tenant, ingest refuses a write that would take the tenant past the effective limit, and the sender receives a refusal rather than an accepted event. Reads, queries and exports stay usable while writes are refused, and lowering a limit never deletes data that is already stored.
The refusal is measured against the tenant's total attributed storage — the tenant-owned database rows plus the raw and archived objects held for it. The Storage screen shows that figure as Storage used, and Remaining and the used percentage are measured on it. Data retained is a different number: the logical size of the tenant's retained events, before compression and without the database or object overhead. The two can differ in either direction, so read Storage used as the figure the limit acts on and Data retained as what is held.
The measurement has to be current. If the platform's storage accounting for your tenant is older than its freshness budget, the hard limit fails closed and writes are refused until the next measurement completes. That is deliberate: the platform does not enforce a limit against accounting it cannot trust. A tenant the platform has never measured is allowed through until the first measurement.
If writes are refused and you are within your intended usage, the recovery is to raise the limit, export what you need, or reduce what is retained. Retrying the same write unchanged will keep being refused.
Inspect and export accepted usage
- Open Admin → Usage. Read the displayed start/end time and source watermark before comparing totals.
- Check Complete windows, Incomplete windows, and Provisional data. A provisional or incomplete window can still change.
- Inspect each row's Account, Data space, Meter, Window, Quantity, and State. Events accepted counts accepted events; Data accepted reports accepted data volume. Read the displayed unit.
- Compare the Window totals for the same account, data space and meter.
- Choose Export CSV or Export JSON. The browser downloads
usage-v1.csvorusage-v1.json; check the account, window and states in the exported data before using it in a reconciliation.
For example, compare an Events accepted total with a sender's accepted results only for the matching window and account. Do not equate accepted usage with Events retained: retention and deletion can reduce what remains stored.
No usage window available means usage has not appeared after processing. No accepted usage in this window means the loaded range has no accepted usage rows. Neither message proves that a different date range is empty.
Revoke only when new writes should stop
Revocation is a separate action from inspecting usage or deleting data.
- In Storage Locations, choose Revoke on the exact location row.
- Read Revoke storage location. New writes will fail; historical event-segment references remain resolvable.
- Type the exact location code shown in the confirmation, then choose Revoke location.
- Reload the list and inspect the result and recent storage history. A closed confirmation alone is not proof that the request succeeded.
Revoking a location is not an erasure request and does not reclaim stored data. Use the appropriate retention or privacy workflow for those tasks.
When something fails
If locations, usage or history cannot load, retain the visible error and retry after access returns. A failed export shows Usage export failed; retry the export after resolving that error rather than treating a missing file as an empty result. For access denial, have your tenant administrator check your role and account access.