Files
netbird/proxy/internal/middleware/registry.go
Maycon Santos b416063bcc [management,proxy] Agent network: per-account LLM gateway (policy, metering, multi-provider) (#6555)
* [agent-network] Shared proto, OpenAPI schema, and generated types

* [agent-network] Management: store, manager, synthesizer, policy engine, provider catalog, HTTP/gRPC API

Adds the account-scoped agent-network module: provider/policy/budget CRUD and
store, the reverse-proxy service synthesizer, policy selection + limit
enforcement, the provider catalog (incl. Vertex AI and AWS Bedrock entries),
and the management HTTP + proxy gRPC surfaces.

* [management] Fix agent-network proxy-peer fan-out on affected-peer recompute

The affected-peers resolver loaded only persisted reverse-proxy services, but
agent-network services are synthesized on demand and never persisted. As a
result the embedded proxy peer was never folded into the affected set when a
client's group changed, so the proxy received no network-map update for a newly
authorised client and rejected its handshake until a full resync (restart).

loadProxyServices now merges the synthesized agent-network services (injected
via a registration hook to avoid an import cycle), so proxy peers learn newly
authorised clients immediately.

* [proxy] Reverse-proxy middleware framework, chain, and request plumbing

The per-target middleware chain (slots, dispatcher, mutation gate, metadata
merger), body capture, access-log terminal sink, and the proxy wiring that
builds + runs chains for synthesized agent-network services.

* [proxy] LLM parsers, pricing, and builtin middlewares (OpenAI, Anthropic, Vertex AI, AWS Bedrock)

Request/response parsers and SSE/event-stream metering, the embedded pricing
table, and the builtin middleware set: request parser, router, policy
limit-check/record, cost meter, guardrail, identity inject, response parser.
Includes the path-routed providers — Google Vertex AI (keyfile:: service-account
OAuth minting) and AWS Bedrock (bearer auth, invoke/converse/streaming, optional
/bedrock prefix) — plus the Models allowlist and unmeterable-publisher deny.

* [proxy] IPv6 in-place apply and TCP accept-loop hardening on netstack listeners

* [agent-network] End-to-end test suite, module docs, and deployment preset

* [agent-network] Fix codespell typos and exclude false positives

- labelgen word pool: vermillion -> vermilion, racoon -> raccoon.
- codespell ignore list: add flate (Go compress/flate package), recordin
  (a test-local identifier), and unparseable (a valid alternative spelling used
  consistently across identifiers + a metadata-value constant).

* [management] Set LastSeen on injected proxy peer in realstack test (MySQL strict-mode)

The injected embedded proxy peer had a PeerStatus with a zero LastSeen, which
serializes to '0000-00-00' and is rejected by MySQL in strict mode (SQLite
tolerates it). Set LastSeen to a valid time so SaveAccount succeeds on both
engines.

* [agent-network] Remove e2e shell-script suite from this branch

The end-to-end shell scripts under scripts/e2e/ are maintained in a separate
testing suite and are not part of this change set.

* [agent-network] Polish module docs: remove internal review scaffolding, fix links, verify diagrams

Strip PR-review framing, commit references, absolute paths, and stale internal
references from the agent-network module docs; fix broken relative links; verify
all diagrams against the current architecture. Remove the internal AI-reviewer
prompt file.

* [management] Refine session expiration handling to support 3-state encoding for SSO deadlines

* [agent-network] Relocate agentnetwork package to internals/modules

Move management/server/agentnetwork (and its catalog/, labelgen/, types/
subpackages) to management/internals/modules/agentnetwork, alongside the
reverse-proxy module, and rewrite all importers. Pure relocation: package names,
the synthesizer + affectedpeers registration hook, and store access (shared
store.Store) are unchanged, so no import cycle is introduced (affectedpeers
still depends only on the agentnetwork/types leaf).

* [agent-network] Co-locate HTTP handlers in the module (RegisterEndpoints)

Move the agent-network HTTP handlers from server/http/handlers/agentnetwork into
the module at internals/modules/agentnetwork/handlers (package handlers) and
rename the entrypoint AddEndpoints -> RegisterEndpoints, matching the
reverse-proxy module convention. Wiring in http/handler.go updated accordingly.
2026-06-27 13:41:00 +02:00

122 lines
3.5 KiB
Go

package middleware
import (
"fmt"
"sync"
)
// Registry maps middleware IDs to their factories. The proxy installs
// a single Registry at boot; concrete middlewares register themselves
// from init() functions inside their own packages so the boot wiring
// only needs an anonymous import.
//
// Registry is safe for concurrent reads after boot. Register / Unregister
// take the write lock; Get and IDs take the read lock.
type Registry struct {
mu sync.RWMutex
factories map[string]Factory
}
// NewRegistry returns an empty registry.
func NewRegistry() *Registry {
return &Registry{factories: make(map[string]Factory)}
}
// Register installs the factory under its ID. Returns an error when an
// ID is already registered — collisions are programmer errors and must
// be visible at boot rather than silently last-write-wins.
func (r *Registry) Register(f Factory) error {
if f == nil {
return fmt.Errorf("middleware registry: nil factory")
}
id := f.ID()
if id == "" {
return fmt.Errorf("middleware registry: factory has empty id")
}
r.mu.Lock()
defer r.mu.Unlock()
if _, exists := r.factories[id]; exists {
return fmt.Errorf("middleware registry: %q already registered", id)
}
r.factories[id] = f
return nil
}
// MustRegister panics on error. Intended for init() registration so
// duplicate IDs surface at startup.
func (r *Registry) MustRegister(f Factory) {
if err := r.Register(f); err != nil {
panic(err)
}
}
// Get returns the factory for id, or nil when no factory is
// registered.
func (r *Registry) Get(id string) Factory {
r.mu.RLock()
defer r.mu.RUnlock()
return r.factories[id]
}
// IDs returns the registered IDs in unspecified order. Used by the
// management translator to reject specs that reference unknown IDs at
// apply time.
func (r *Registry) IDs() []string {
r.mu.RLock()
defer r.mu.RUnlock()
out := make([]string, 0, len(r.factories))
for id := range r.factories {
out = append(out, id)
}
return out
}
// IsKnown reports whether id has a registered factory.
func (r *Registry) IsKnown(id string) bool {
return r.Get(id) != nil
}
// Resolver wraps a Registry and produces a configured Middleware
// instance from a Spec. The Manager uses this during chain build.
type Resolver struct {
registry *Registry
}
// NewResolver returns a resolver backed by the registry.
func NewResolver(registry *Registry) *Resolver {
if registry == nil {
registry = NewRegistry()
}
return &Resolver{registry: registry}
}
// Resolve builds a Middleware instance and merges runtime-only fields
// (version, accepted content types, metadata key allowlist, mutation
// support) onto the spec.
//
// Return semantics:
// - (mw, mergedSpec, nil): instance built, include in chain.
// - (nil, spec, nil): id not registered; silently skip.
// - (nil, spec, err): factory rejected the config (logged + counted
// by Manager, other middlewares still bind).
func (r *Resolver) Resolve(spec Spec) (Middleware, Spec, error) {
f := r.registry.Get(spec.ID)
if f == nil {
return nil, spec, nil
}
mw, err := f.New(spec.RawConfig)
if err != nil {
return nil, spec, fmt.Errorf("middleware %s factory: %w", spec.ID, err)
}
if mw.Slot() != spec.Slot {
_ = mw.Close()
return nil, spec, fmt.Errorf("middleware %s slot mismatch: spec=%d impl=%d", spec.ID, spec.Slot, mw.Slot())
}
merged := spec
merged.Version = mw.Version()
merged.MetadataKeys = append([]string(nil), mw.MetadataKeys()...)
merged.AcceptedContentTypes = append([]string(nil), mw.AcceptedContentTypes()...)
merged.MutationsSupported = mw.MutationsSupported()
return mw, merged, nil
}