Delete vector records

POST /v1/tenants/{tenant}/vector-indexes/{index}/records/delete

Remove selected vector records deliberatelyID deletion, filtered deletion and clearing a namespace have different outcomes.

Choose one deletion scopeUse IDs, a non-empty filter or all records.
Review the requestFor delete-all, confirmation must match the namespace.
Inspect counts or job progressA filtered job can continue in the background.

Before you start

Use your database bearer token.

Connection setup

Choose one selector: IDs, a metadata filter, or the whole namespace. The example deletes only v1.

Path parameters

Field	Type	Required
`tenant`	`string`	Required
`index`	`string`	Required

Headers

Field	Type	Required	Description
`x-oc-db-user`	`string`	Optional	Acting database user (role, grant and row-policy checks). Send with x-oc-db-key.
`x-oc-db-key`	`string`	Optional	API key of the x-oc-db-user user.

Request body

application/json

Limits and accepted values

Three forms;
exactly one of ids, a filter (filter_expr or filter, at most one) or all: true, otherwise 400 exactly_one_selector.
By ids: ids holds 1-1,000 distinct ids of at most 1,024 bytes (dry_run: true refused with 400).
By filter: an empty filter is refused (400 filter_empty_for_mutation);
dry_run: true counts the matches and writes nothing;
a filtered job deletes at most 1,000,000 records, then ends limit_reached;
on a namespace holding more than 100,000 records it is refused 501 not_implemented (details.reason = delete_by_filter_namespace_too_large, a provisional bound).
One running and 100 retained filtered jobs per namespace;
beyond either, 429 job_limit_reached (retryable;
details.bound, details.limit).
A filtered job needs a key of at least 32 bytes (403 key_scope_insufficient).
The job is read at GET .../jobs/{job_id}
and stopped at POST .../jobs/{job_id}/cancel.
Delete-all: all: true with confirm equal to namespace;
dry_run: true returns the count without deleting.

Field	Type	Required
`all`	`boolean`	Optional
`confirm`	`string`	Optional
`dry_run`	`boolean`	Optional
`filter`	`object`	Optional
`filter_expr`	`object`	Optional
`ids`	`string[]`	Optional

Fields in ids

Field	Type	Required
`ids`	`string`	Optional

Field	Type	Required
`namespace`	`string`	Required

By-ID results report deleted and missing counts. A filtered delete can instead start a job; track that job to learn how much work completed. For delete-all, confirm must match the namespace exactly.

Response statuses

HTTP status reference

Status	What to do
200	OK
202	Accepted
400	Correct the request before retrying.
401	Check the credential and Authorization header.
402	Check feature access or available credits.
403	Check tenant access, roles and permissions.
404	Check the resource ID and API route.
409	Resolve the conflict before retrying.
413	Split or reduce the request body.
415	Send the required Content-Type.
422	Correct the values or unsupported combination.
429	Honor Retry-After when present; back off.
500	Check the outcome before retrying a write.
501	Check deployment support for this operation.
503	Honor Retry-After when present; check write outcomes.

Request example

curl --include --request POST "${OC_BASE_URL}/v1/tenants/${OC_TENANT}/vector-indexes/demo-search/records/delete" \
  --header "Authorization: Bearer ${OC_TOKEN}" \
  --header "Content-Type: application/json" \
  --data-binary '{
  "namespace": "demo",
  "ids": [
    "v1"
  ]
}'

Response example

Examples use synthetic values to show the response format.

Response status

{
  "deleted": 1,
  "missing": 0,
  "receipt": {
    "tenant": "your-tenant",
    "index_id": "22222222-2222-4222-8222-222222222222",
    "namespace_id": "33333333-3333-4333-8333-333333333333",
    "incarnation_id": "66666666-6666-4666-8666-666666666666",
    "restore_epoch": {
      "counter": 0,
      "nonce_prefix": "<opaque-restore-prefix>"
    },
    "position": {
      "domain": "wal",
      "value": "1042:0"
    },
    "durability_scope": "local_fsync",
    "records": [
      {
        "id": "v1",
        "state": "durable"
      }
    ]
  }
}

{
  "job_id": "55555555-5555-4555-8555-555555555555",
  "receipt": {
    "tenant": "your-tenant",
    "index_id": "22222222-2222-4222-8222-222222222222",
    "namespace_id": "33333333-3333-4333-8333-333333333333",
    "incarnation_id": "66666666-6666-4666-8666-666666666666",
    "restore_epoch": {
      "counter": 0,
      "nonce_prefix": "<opaque-restore-prefix>"
    },
    "position": {
      "domain": "wal",
      "value": "1042:0"
    },
    "durability_scope": "local_fsync",
    "records": []
  }
}

202 illustrates a filter-based mutation; the displayed id-based request returns 200 instead.

{
  "error": {
    "code": "exactly_one_selector",
    "message": "Supply exactly one record selector.",
    "status": 400,
    "retryable": false
  }
}

Illustrative message; code and envelope follow the documented vector-index contract.

Correct the request before retrying.

{
  "error": {
    "code": "job_limit_reached",
    "message": "The namespace has reached its concurrent job limit.",
    "status": 429,
    "retryable": true,
    "details": {
      "bound": "running",
      "limit": 1
    }
  }
}

Illustrative message; code and envelope follow the documented vector-index contract.

Honor Retry-After when present; back off.

Download the OpenAPI specification for the complete response schemas.

Before you startCopy Anchor Link

Path parametersCopy Anchor Link

HeadersCopy Anchor Link

Request bodyCopy Anchor Link

Read the resultCopy Anchor Link

Response statusesCopy Anchor Link

Before you start

Path parameters

Headers

Request body

Read the result

Response statuses