Files
milvus/internal/util/function/validator/alter_whitelist.go
T
5def5ced4a enhance: enforce function-field binding principle for function DDL (#51360)
## What

Enforce a function–field binding principle for function DDL (add / drop
/ alter function on an existing collection), so a function and its
output field stay coupled and already-stored vectors can never be
silently invalidated by DDL.

BM25 and MinHash follow the strict binding: add/drop only via the
unified `add_function_field` / `drop_function_field`, output field
always coupled. **TextEmbedding is a temporary carve-out** — its legacy
`AddCollectionFunction` / `DropCollectionFunction` paths are retained,
so attach-over-existing-field and detach are still possible for it,
because `add_function_field` embedding backfill is not yet implemented.
It will be folded into the unified path in a follow-up. So the "always
coupled" invariant currently holds for BM25/MinHash, not TextEmbedding.

## Changes

- **AlterCollectionSchema invariant**: reject standalone add-function (a
function must be added together with its new output field), reject
detaching a function without dropping its output field, and reject
function cascade (a new function's input being another function's
output).
- **Legacy RPCs**: `AddCollectionFunction` / `DropCollectionFunction`
are rejected for BM25/MinHash (must use `add_function_field` /
`drop_function_field`) and retained only for TextEmbedding (see
carve-out). `AlterCollectionFunction` is kept and gains the whitelist
below. In the alter and legacy-add paths, function field IDs are always
re-derived from field names (never trusted from the request), so a
request cannot inject an unrelated field ID that a later
`drop_function_field` would delete.
- **alter_function whitelist**: only connection/runtime params may
change (TextEmbedding: `url`, `credential`, `timeout_ms`,
`max_client_batch_size`, `region`, `location`, `projectid`, `user`);
BM25/MinHash have no alterable params. Function identity (type, name,
input/output fields) and output-shaping params (`dim`, `model_name`,
`endpoint` — the TEI model identity, `normalize`, `truncate*`, prompts,
...) are immutable. Params are normalized (keys lowercased, duplicate
keys rejected) so a crafted duplicate key cannot bypass the diff.
- **drop_function_field**: allow any output-producing function (dropping
needs no backfill), removing the previous BM25/MinHash-only restriction.

## Not in scope / follow-up

- Extending `add_function_field` to TextEmbedding (post-creation add
would require backfilling every existing row through the external
embedding model) — deferred; folding the TextEmbedding legacy paths into
the unified binding follows this.
- MinHash input-field analyzer immutability lives on a different DDL
path (`AlterCollectionField`); tracked as a separate follow-up.

## Breaking changes

- `DropCollectionFunction` on a **MinHash** function (detach — remove
the function but keep its output field) is no longer supported; use
`drop_function_field` instead (drops the function together with its
output field). If a released version supported MinHash detach, migrate
any such call to `drop_function_field`.

issue: #51348

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Signed-off-by: MrPresent-Han <chun.han@gmail.com>
Co-authored-by: MrPresent-Han <chun.han@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-18 05:04:39 +08:00

127 lines
5.1 KiB
Go

// Licensed to the LF AI & Data foundation under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package validator
import (
"slices"
"strings"
"github.com/milvus-io/milvus-proto/go-api/v3/commonpb"
"github.com/milvus-io/milvus-proto/go-api/v3/schemapb"
"github.com/milvus-io/milvus/internal/util/function/models"
"github.com/milvus-io/milvus/pkg/v3/util/merr"
"github.com/milvus-io/milvus/pkg/v3/util/typeutil"
)
// alterableFunctionParams lists, per function type, the param keys that
// alter_function may change. Only pure connection/auth/perf knobs are allowed —
// params that never change the produced vector for a given input. Any param that
// can change the output vector's semantics or shape is immutable, because altering
// it on an already-backfilled output field would silently mix incompatible vectors
// in the same field. A function type absent from this map (BM25, MinHash) has no
// alterable params.
//
// Immutable (semantic) params, by first principle "same input must keep producing a
// compatible vector":
// - dim / model_name / provider / normalization / prompts: change the model or its
// output shape.
// - endpoint: for TEI the endpoint IS the model's identity, so repointing it can
// silently swap to a different model.
// - truncate / truncation_direction: for an over-length input they change which
// tokens are embedded (truncate on/off, head vs tail), so the produced vector
// differs — altering them mixes old and new vector semantics in one field.
//
// url stays alterable because model_name providers keep their identity in model_name
// and use url only as a relocatable API host.
var alterableFunctionParams = map[schemapb.FunctionType]typeutil.Set[string]{
schemapb.FunctionType_TextEmbedding: typeutil.NewSet(
models.URLParamKey,
models.CredentialParamKey,
models.TimeoutMsParamKey,
models.MaxClientBatchSizeParamKey,
models.RegionParamKey,
models.LocationParamKey,
models.ProjectIDParamKey,
models.UserParamKey,
),
}
// CheckFunctionAlterAllowed rejects an alter_function request that changes
// anything other than whitelisted params. The function identity (type, name,
// input/output fields) is immutable; only connection/runtime params may change.
func CheckFunctionAlterAllowed(oldFn, newFn *schemapb.FunctionSchema) error {
if oldFn.GetType() != newFn.GetType() {
return merr.WrapErrParameterInvalidMsg("function type cannot be altered: function %s", oldFn.GetName())
}
if oldFn.GetName() != newFn.GetName() {
return merr.WrapErrParameterInvalidMsg("function name cannot be altered: %s -> %s", oldFn.GetName(), newFn.GetName())
}
if !slices.Equal(oldFn.GetInputFieldNames(), newFn.GetInputFieldNames()) {
return merr.WrapErrParameterInvalidMsg("function input fields cannot be altered: function %s", oldFn.GetName())
}
if !slices.Equal(oldFn.GetOutputFieldNames(), newFn.GetOutputFieldNames()) {
return merr.WrapErrParameterInvalidMsg("function output fields cannot be altered: function %s", oldFn.GetName())
}
whitelist := alterableFunctionParams[oldFn.GetType()]
if len(whitelist) == 0 {
return merr.WrapErrParameterInvalidMsg("function type %s has no alterable params", oldFn.GetType().String())
}
oldParams, err := normalizeParams(oldFn.GetParams())
if err != nil {
return err
}
newParams, err := normalizeParams(newFn.GetParams())
if err != nil {
return err
}
changed := typeutil.NewSet[string]()
for k, ov := range oldParams {
if nv, ok := newParams[k]; !ok || nv != ov {
changed.Insert(k)
}
}
for k := range newParams {
if _, ok := oldParams[k]; !ok {
changed.Insert(k)
}
}
for k := range changed {
if !whitelist.Contain(k) {
return merr.WrapErrParameterInvalidMsg(
"function param %q cannot be altered for function %s; only connection/runtime params may be changed", k, oldFn.GetName())
}
}
return nil
}
// normalizeParams lowercases param keys (matching how providers read them, e.g.
// getProvider) and rejects duplicate keys, so a crafted duplicate key cannot slip
// a non-whitelisted change past the diff while the runtime binds to a different
// occurrence.
func normalizeParams(params []*commonpb.KeyValuePair) (map[string]string, error) {
m := make(map[string]string, len(params))
for _, p := range params {
k := strings.ToLower(p.GetKey())
if _, dup := m[k]; dup {
return nil, merr.WrapErrParameterInvalidMsg("duplicate function param key %q", p.GetKey())
}
m[k] = p.GetValue()
}
return m, nil
}