mirror of
https://github.com/netbirdio/netbird.git
synced 2026-09-23 07:09:08 +02:00
[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.
This commit is contained in:
Maycon Santos
@@ -37,6 +37,7 @@ import (
|
||||
rpservice "github.com/netbirdio/netbird/management/internals/modules/reverseproxy/service"
|
||||
"github.com/netbirdio/netbird/management/internals/modules/zones"
|
||||
"github.com/netbirdio/netbird/management/internals/modules/zones/records"
|
||||
agentNetworkTypes "github.com/netbirdio/netbird/management/internals/modules/agentnetwork/types"
|
||||
resourceTypes "github.com/netbirdio/netbird/management/server/networks/resources/types"
|
||||
routerTypes "github.com/netbirdio/netbird/management/server/networks/routers/types"
|
||||
networkTypes "github.com/netbirdio/netbird/management/server/networks/types"
|
||||
@@ -137,6 +138,10 @@ func NewSqlStore(ctx context.Context, db *gorm.DB, storeEngine types.Engine, met
|
||||
&networkTypes.Network{}, &routerTypes.NetworkRouter{}, &resourceTypes.NetworkResource{}, &types.AccountOnboarding{},
|
||||
&types.Job{}, &zones.Zone{}, &records.Record{}, &types.UserInviteRecord{}, &rpservice.Service{}, &rpservice.Target{}, &domain.Domain{},
|
||||
&accesslogs.AccessLogEntry{}, &proxy.Proxy{},
|
||||
&agentNetworkTypes.Provider{}, &agentNetworkTypes.Policy{}, &agentNetworkTypes.Guardrail{}, &agentNetworkTypes.Settings{},
|
||||
&agentNetworkTypes.Consumption{}, &agentNetworkTypes.AccountBudgetRule{},
|
||||
&agentNetworkTypes.AgentNetworkAccessLog{}, &agentNetworkTypes.AgentNetworkAccessLogGroup{},
|
||||
&agentNetworkTypes.AgentNetworkUsage{}, &agentNetworkTypes.AgentNetworkUsageGroup{},
|
||||
)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("auto migratePreAuto: %w", err)
|
||||
@@ -5573,6 +5578,255 @@ func (s *SqlStore) CreateAccessLog(ctx context.Context, logEntry *accesslogs.Acc
|
||||
return nil
|
||||
}
|
||||
|
||||
// CreateAgentNetworkAccessLog persists a flattened agent-network access-log
|
||||
// entry together with its authorising-group child rows in a single
|
||||
// transaction.
|
||||
func (s *SqlStore) CreateAgentNetworkAccessLog(ctx context.Context, entry *agentNetworkTypes.AgentNetworkAccessLog, groups []agentNetworkTypes.AgentNetworkAccessLogGroup) error {
|
||||
err := s.db.Transaction(func(tx *gorm.DB) error {
|
||||
// Idempotent on the log id / (log_id, group_id) so a proxy resend of the
|
||||
// same entry can't fail the request.
|
||||
if err := tx.Clauses(clause.OnConflict{DoNothing: true}).Create(entry).Error; err != nil {
|
||||
return err
|
||||
}
|
||||
if len(groups) > 0 {
|
||||
if err := tx.Clauses(clause.OnConflict{DoNothing: true}).Create(&groups).Error; err != nil {
|
||||
return err
|
||||
}
|
||||
}
|
||||
return nil
|
||||
})
|
||||
if err != nil {
|
||||
log.WithContext(ctx).WithFields(log.Fields{
|
||||
"account_id": entry.AccountID,
|
||||
"service_id": entry.ServiceID,
|
||||
"model": entry.Model,
|
||||
}).Errorf("failed to create agent-network access log entry in store: %v", err)
|
||||
return status.Errorf(status.Internal, "failed to create agent-network access log entry in store")
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// CreateAgentNetworkUsage persists a stripped agent-network usage record
|
||||
// together with its authorising-group child rows in a single transaction.
|
||||
func (s *SqlStore) CreateAgentNetworkUsage(ctx context.Context, usage *agentNetworkTypes.AgentNetworkUsage, groups []agentNetworkTypes.AgentNetworkUsageGroup) error {
|
||||
err := s.db.Transaction(func(tx *gorm.DB) error {
|
||||
// Idempotent on the usage id / (usage_id, group_id) so a proxy resend of
|
||||
// the same entry can't fail the request.
|
||||
if err := tx.Clauses(clause.OnConflict{DoNothing: true}).Create(usage).Error; err != nil {
|
||||
return err
|
||||
}
|
||||
if len(groups) > 0 {
|
||||
if err := tx.Clauses(clause.OnConflict{DoNothing: true}).Create(&groups).Error; err != nil {
|
||||
return err
|
||||
}
|
||||
}
|
||||
return nil
|
||||
})
|
||||
if err != nil {
|
||||
log.WithContext(ctx).WithFields(log.Fields{
|
||||
"account_id": usage.AccountID,
|
||||
"model": usage.Model,
|
||||
}).Errorf("failed to create agent-network usage record in store: %v", err)
|
||||
return status.Errorf(status.Internal, "failed to create agent-network usage record in store")
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// DeleteOldAgentNetworkAccessLogs deletes an account's access-log rows (and
|
||||
// their authorising-group child rows) older than the cutoff. Usage records are
|
||||
// untouched — they are the long-term aggregate. Returns the number of log rows
|
||||
// deleted.
|
||||
func (s *SqlStore) DeleteOldAgentNetworkAccessLogs(ctx context.Context, accountID string, olderThan time.Time) (int64, error) {
|
||||
var deleted int64
|
||||
err := s.db.Transaction(func(tx *gorm.DB) error {
|
||||
// Remove group child rows for the soon-to-be-deleted logs first.
|
||||
if err := tx.Exec(
|
||||
"DELETE FROM agent_network_access_log_group WHERE account_id = ? AND log_id IN (SELECT id FROM agent_network_access_log WHERE account_id = ? AND timestamp < ?)",
|
||||
accountID, accountID, olderThan,
|
||||
).Error; err != nil {
|
||||
return err
|
||||
}
|
||||
res := tx.Where("account_id = ? AND timestamp < ?", accountID, olderThan).
|
||||
Delete(&agentNetworkTypes.AgentNetworkAccessLog{})
|
||||
if res.Error != nil {
|
||||
return res.Error
|
||||
}
|
||||
deleted = res.RowsAffected
|
||||
return nil
|
||||
})
|
||||
if err != nil {
|
||||
log.WithContext(ctx).Errorf("failed to delete old agent-network access logs for account %s: %v", accountID, err)
|
||||
return 0, status.Errorf(status.Internal, "failed to delete old agent-network access logs")
|
||||
}
|
||||
return deleted, nil
|
||||
}
|
||||
|
||||
// GetAgentNetworkUsageRows returns the stripped usage rows for an account that
|
||||
// match the filter (date / user / group / provider / model). Aggregation into
|
||||
// time buckets happens in the manager so granularities stay engine-portable.
|
||||
func (s *SqlStore) GetAgentNetworkUsageRows(ctx context.Context, lockStrength LockingStrength, accountID string, filter agentNetworkTypes.AgentNetworkAccessLogFilter) ([]*agentNetworkTypes.AgentNetworkUsage, error) {
|
||||
var rows []*agentNetworkTypes.AgentNetworkUsage
|
||||
|
||||
query := s.applyAgentNetworkUsageFilters(
|
||||
s.db.Where(accountIDCondition, accountID),
|
||||
filter,
|
||||
).Order("timestamp ASC")
|
||||
|
||||
if lockStrength != LockingStrengthNone {
|
||||
query = query.Clauses(clause.Locking{Strength: string(lockStrength)})
|
||||
}
|
||||
|
||||
if err := query.Find(&rows).Error; err != nil {
|
||||
log.WithContext(ctx).Errorf("failed to get agent-network usage rows from store: %v", err)
|
||||
return nil, status.Errorf(status.Internal, "failed to get agent-network usage rows from store")
|
||||
}
|
||||
return rows, nil
|
||||
}
|
||||
|
||||
// applyAgentNetworkUsageFilters applies the shared access-log filter's
|
||||
// date/user/group/provider/model conditions to a usage-table query. Pagination,
|
||||
// sort and free-text search are ignored — the overview is an aggregate.
|
||||
func (s *SqlStore) applyAgentNetworkUsageFilters(query *gorm.DB, filter agentNetworkTypes.AgentNetworkAccessLogFilter) *gorm.DB {
|
||||
if filter.UserID != nil {
|
||||
query = query.Where("user_id = ?", *filter.UserID)
|
||||
}
|
||||
if filter.SessionID != nil {
|
||||
query = query.Where("session_id = ?", *filter.SessionID)
|
||||
}
|
||||
if len(filter.ProviderIDs) > 0 {
|
||||
query = query.Where("resolved_provider_id IN ?", filter.ProviderIDs)
|
||||
}
|
||||
if len(filter.Models) > 0 {
|
||||
query = query.Where("model IN ?", filter.Models)
|
||||
}
|
||||
if len(filter.GroupIDs) > 0 {
|
||||
query = query.Where(
|
||||
"id IN (SELECT usage_id FROM agent_network_request_usage_group WHERE group_id IN ?)",
|
||||
filter.GroupIDs,
|
||||
)
|
||||
}
|
||||
if filter.StartDate != nil {
|
||||
query = query.Where("timestamp >= ?", *filter.StartDate)
|
||||
}
|
||||
if filter.EndDate != nil {
|
||||
query = query.Where("timestamp <= ?", *filter.EndDate)
|
||||
}
|
||||
return query
|
||||
}
|
||||
|
||||
// GetAgentNetworkAccessLogs retrieves flattened agent-network access logs for
|
||||
// an account with server-side pagination, filtering and sorting. Authorising
|
||||
// group ids are hydrated from the group child table for the returned page.
|
||||
func (s *SqlStore) GetAgentNetworkAccessLogs(ctx context.Context, lockStrength LockingStrength, accountID string, filter agentNetworkTypes.AgentNetworkAccessLogFilter) ([]*agentNetworkTypes.AgentNetworkAccessLog, int64, error) {
|
||||
var logs []*agentNetworkTypes.AgentNetworkAccessLog
|
||||
var totalCount int64
|
||||
|
||||
countQuery := s.applyAgentNetworkAccessLogFilters(
|
||||
s.db.Model(&agentNetworkTypes.AgentNetworkAccessLog{}).Where(accountIDCondition, accountID),
|
||||
filter,
|
||||
)
|
||||
if err := countQuery.Count(&totalCount).Error; err != nil {
|
||||
log.WithContext(ctx).Errorf("failed to count agent-network access logs: %v", err)
|
||||
return nil, 0, status.Errorf(status.Internal, "failed to count agent-network access logs")
|
||||
}
|
||||
|
||||
query := s.applyAgentNetworkAccessLogFilters(
|
||||
s.db.Where(accountIDCondition, accountID),
|
||||
filter,
|
||||
).
|
||||
Order(filter.GetSortColumn() + " " + filter.GetSortOrder()).
|
||||
Limit(filter.GetLimit()).
|
||||
Offset(filter.GetOffset())
|
||||
|
||||
if lockStrength != LockingStrengthNone {
|
||||
query = query.Clauses(clause.Locking{Strength: string(lockStrength)})
|
||||
}
|
||||
|
||||
if err := query.Find(&logs).Error; err != nil {
|
||||
log.WithContext(ctx).Errorf("failed to get agent-network access logs from store: %v", err)
|
||||
return nil, 0, status.Errorf(status.Internal, "failed to get agent-network access logs from store")
|
||||
}
|
||||
|
||||
if err := s.hydrateAgentNetworkAccessLogGroups(ctx, accountID, logs); err != nil {
|
||||
return nil, 0, err
|
||||
}
|
||||
|
||||
return logs, totalCount, nil
|
||||
}
|
||||
|
||||
// applyAgentNetworkAccessLogFilters applies the filter conditions to a query.
|
||||
func (s *SqlStore) applyAgentNetworkAccessLogFilters(query *gorm.DB, filter agentNetworkTypes.AgentNetworkAccessLogFilter) *gorm.DB {
|
||||
if filter.Search != nil {
|
||||
p := "%" + *filter.Search + "%"
|
||||
query = query.Where(
|
||||
"id LIKE ? OR host LIKE ? OR path LIKE ? OR model LIKE ? OR user_id IN (SELECT id FROM users WHERE email LIKE ? OR name LIKE ?)",
|
||||
p, p, p, p, p, p,
|
||||
)
|
||||
}
|
||||
if filter.UserID != nil {
|
||||
query = query.Where("user_id = ?", *filter.UserID)
|
||||
}
|
||||
if filter.SessionID != nil {
|
||||
query = query.Where("session_id = ?", *filter.SessionID)
|
||||
}
|
||||
if filter.Decision != nil {
|
||||
query = query.Where("decision = ?", *filter.Decision)
|
||||
}
|
||||
if filter.PathPrefix != nil {
|
||||
query = query.Where("path LIKE ?", *filter.PathPrefix+"%")
|
||||
}
|
||||
if len(filter.ProviderIDs) > 0 {
|
||||
query = query.Where("resolved_provider_id IN ?", filter.ProviderIDs)
|
||||
}
|
||||
if len(filter.Models) > 0 {
|
||||
query = query.Where("model IN ?", filter.Models)
|
||||
}
|
||||
if len(filter.GroupIDs) > 0 {
|
||||
query = query.Where(
|
||||
"id IN (SELECT log_id FROM agent_network_access_log_group WHERE group_id IN ?)",
|
||||
filter.GroupIDs,
|
||||
)
|
||||
}
|
||||
if filter.StartDate != nil {
|
||||
query = query.Where("timestamp >= ?", *filter.StartDate)
|
||||
}
|
||||
if filter.EndDate != nil {
|
||||
query = query.Where("timestamp <= ?", *filter.EndDate)
|
||||
}
|
||||
return query
|
||||
}
|
||||
|
||||
// hydrateAgentNetworkAccessLogGroups loads the authorising group ids for the
|
||||
// given page of entries and assigns them onto each entry's GroupIDs field.
|
||||
func (s *SqlStore) hydrateAgentNetworkAccessLogGroups(ctx context.Context, accountID string, logs []*agentNetworkTypes.AgentNetworkAccessLog) error {
|
||||
if len(logs) == 0 {
|
||||
return nil
|
||||
}
|
||||
|
||||
ids := make([]string, 0, len(logs))
|
||||
for _, l := range logs {
|
||||
ids = append(ids, l.ID)
|
||||
}
|
||||
|
||||
var rows []agentNetworkTypes.AgentNetworkAccessLogGroup
|
||||
if err := s.db.
|
||||
Where(accountIDCondition, accountID).
|
||||
Where("log_id IN ?", ids).
|
||||
Find(&rows).Error; err != nil {
|
||||
log.WithContext(ctx).Errorf("failed to hydrate agent-network access log groups: %v", err)
|
||||
return status.Errorf(status.Internal, "failed to hydrate agent-network access log groups")
|
||||
}
|
||||
|
||||
byLog := make(map[string][]string, len(logs))
|
||||
for _, r := range rows {
|
||||
byLog[r.LogID] = append(byLog[r.LogID], r.GroupID)
|
||||
}
|
||||
for _, l := range logs {
|
||||
l.GroupIDs = byLog[l.ID]
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// GetAccountAccessLogs retrieves access logs for a given account with pagination and filtering
|
||||
func (s *SqlStore) GetAccountAccessLogs(ctx context.Context, lockStrength LockingStrength, accountID string, filter accesslogs.AccessLogFilter) ([]*accesslogs.AccessLogEntry, int64, error) {
|
||||
var logs []*accesslogs.AccessLogEntry
|
||||
|
||||
Reference in New Issue
Block a user