Files
zeekay b61441f7d7 Volume: node-local persistent volumes, under our own name
A fork of Rancher's local-path-provisioner at v0.0.36. The behaviour is
unchanged and deliberately so — a directory on the node a pod lands on is
the right thing for a disk. What changes is identity.

The provisioner answers to hanzo.ai/volume rather than rancher.io/local-path.
That is not branding: two provisioners sharing one name race for the same
claims, and a stock k3s carries the upstream one, so a cluster running both
would settle our storage by chance. Data lives at /var/lib/hanzo/volume, so
reading a Hanzo cluster does not mean learning another vendor's directory.

Apache-2.0 as published, unaltered, with Rancher's attribution in NOTICE.

The README says what this is not: durability. A volume here dies with its
node. That is hanzoai/replicate's question, one layer up, and keeping the
two apart is what lets the disk stay this simple.
2026-09-07 17:30:01 -07:00

210 lines
5.3 KiB
Go

/*
Copyright 2019 The Kubernetes Authors.
Licensed 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 value
import (
"reflect"
)
type mapReflect struct {
valueReflect
}
func (r mapReflect) Length() int {
val := r.Value
return val.Len()
}
func (r mapReflect) Empty() bool {
val := r.Value
return val.Len() == 0
}
func (r mapReflect) Get(key string) (Value, bool) {
return r.GetUsing(HeapAllocator, key)
}
func (r mapReflect) GetUsing(a Allocator, key string) (Value, bool) {
k, v, ok := r.get(key)
if !ok {
return nil, false
}
return a.allocValueReflect().mustReuse(v, nil, &r.Value, &k), true
}
func (r mapReflect) get(k string) (key, value reflect.Value, ok bool) {
mapKey := r.toMapKey(k)
val := r.Value.MapIndex(mapKey)
return mapKey, val, val.IsValid() && val != reflect.Value{}
}
func (r mapReflect) Has(key string) bool {
var val reflect.Value
val = r.Value.MapIndex(r.toMapKey(key))
if !val.IsValid() {
return false
}
return val != reflect.Value{}
}
func (r mapReflect) Set(key string, val Value) {
r.Value.SetMapIndex(r.toMapKey(key), reflect.ValueOf(val.Unstructured()))
}
func (r mapReflect) Delete(key string) {
val := r.Value
val.SetMapIndex(r.toMapKey(key), reflect.Value{})
}
// TODO: Do we need to support types that implement json.Marshaler and are used as string keys?
func (r mapReflect) toMapKey(key string) reflect.Value {
val := r.Value
return reflect.ValueOf(key).Convert(val.Type().Key())
}
func (r mapReflect) Iterate(fn func(string, Value) bool) bool {
return r.IterateUsing(HeapAllocator, fn)
}
func (r mapReflect) IterateUsing(a Allocator, fn func(string, Value) bool) bool {
if r.Value.Len() == 0 {
return true
}
v := a.allocValueReflect()
defer a.Free(v)
return eachMapEntry(r.Value, func(e *TypeReflectCacheEntry, key reflect.Value, value reflect.Value) bool {
return fn(key.String(), v.mustReuse(value, e, &r.Value, &key))
})
}
func eachMapEntry(val reflect.Value, fn func(*TypeReflectCacheEntry, reflect.Value, reflect.Value) bool) bool {
iter := val.MapRange()
entry := TypeReflectEntryOf(val.Type().Elem())
for iter.Next() {
next := iter.Value()
if !next.IsValid() {
continue
}
if !fn(entry, iter.Key(), next) {
return false
}
}
return true
}
func (r mapReflect) Unstructured() interface{} {
result := make(map[string]interface{}, r.Length())
r.Iterate(func(s string, value Value) bool {
result[s] = value.Unstructured()
return true
})
return result
}
func (r mapReflect) Equals(m Map) bool {
return r.EqualsUsing(HeapAllocator, m)
}
func (r mapReflect) EqualsUsing(a Allocator, m Map) bool {
lhsLength := r.Length()
rhsLength := m.Length()
if lhsLength != rhsLength {
return false
}
if lhsLength == 0 {
return true
}
vr := a.allocValueReflect()
defer a.Free(vr)
entry := TypeReflectEntryOf(r.Value.Type().Elem())
return m.Iterate(func(key string, value Value) bool {
_, lhsVal, ok := r.get(key)
if !ok {
return false
}
return EqualsUsing(a, vr.mustReuse(lhsVal, entry, nil, nil), value)
})
}
func (r mapReflect) Zip(other Map, order MapTraverseOrder, fn func(key string, lhs, rhs Value) bool) bool {
return r.ZipUsing(HeapAllocator, other, order, fn)
}
func (r mapReflect) ZipUsing(a Allocator, other Map, order MapTraverseOrder, fn func(key string, lhs, rhs Value) bool) bool {
if otherMapReflect, ok := other.(*mapReflect); ok && order == Unordered {
return r.unorderedReflectZip(a, otherMapReflect, fn)
}
return defaultMapZip(a, &r, other, order, fn)
}
// unorderedReflectZip provides an optimized unordered zip for mapReflect types.
func (r mapReflect) unorderedReflectZip(a Allocator, other *mapReflect, fn func(key string, lhs, rhs Value) bool) bool {
if r.Empty() && (other == nil || other.Empty()) {
return true
}
lhs := r.Value
lhsEntry := TypeReflectEntryOf(lhs.Type().Elem())
// map lookup via reflection is expensive enough that it is better to keep track of visited keys
visited := map[string]struct{}{}
vlhs, vrhs := a.allocValueReflect(), a.allocValueReflect()
defer a.Free(vlhs)
defer a.Free(vrhs)
if other != nil {
rhs := other.Value
rhsEntry := TypeReflectEntryOf(rhs.Type().Elem())
iter := rhs.MapRange()
for iter.Next() {
key := iter.Key()
keyString := key.String()
next := iter.Value()
if !next.IsValid() {
continue
}
rhsVal := vrhs.mustReuse(next, rhsEntry, &rhs, &key)
visited[keyString] = struct{}{}
var lhsVal Value
if _, v, ok := r.get(keyString); ok {
lhsVal = vlhs.mustReuse(v, lhsEntry, &lhs, &key)
}
if !fn(keyString, lhsVal, rhsVal) {
return false
}
}
}
iter := lhs.MapRange()
for iter.Next() {
key := iter.Key()
if _, ok := visited[key.String()]; ok {
continue
}
next := iter.Value()
if !next.IsValid() {
continue
}
if !fn(key.String(), vlhs.mustReuse(next, lhsEntry, &lhs, &key), nil) {
return false
}
}
return true
}