vpp-maglev/internal/grpcapi/server.go

// Copyright (c) 2026, Pim van Pelt <pim@ipng.ch>

package grpcapi

import (
	"context"
	"errors"
	"log/slog"
	"net"
	"sort"

	"google.golang.org/grpc/codes"
	"google.golang.org/grpc/status"

	"git.ipng.ch/ipng/vpp-maglev/internal/checker"
	"git.ipng.ch/ipng/vpp-maglev/internal/config"
	"git.ipng.ch/ipng/vpp-maglev/internal/health"
	"git.ipng.ch/ipng/vpp-maglev/internal/vpp"
)

// Server implements the MaglevServer gRPC interface.
type Server struct {
	UnimplementedMaglevServer
	ctx        context.Context
	checker    *checker.Checker
	logs       *LogBroadcaster
	configPath string
	vppClient  *vpp.Client // nil when VPP integration is disabled
}

// NewServer creates a Server backed by the given Checker. logs may be nil, in
// which case log events are never sent to WatchEvents streams. configPath is
// used by CheckConfig to reload and validate the configuration file on demand.
// vppClient may be nil if VPP integration is disabled. The provided context
// controls the lifetime of streaming RPCs: cancelling it closes all active
// WatchEvents streams so that grpc.Server.GracefulStop can complete.
func NewServer(ctx context.Context, c *checker.Checker, logs *LogBroadcaster, configPath string, vppClient *vpp.Client) *Server {
	return &Server{ctx: ctx, checker: c, logs: logs, configPath: configPath, vppClient: vppClient}
}

// ListFrontends returns the names of all configured frontends.
func (s *Server) ListFrontends(_ context.Context, _ *ListFrontendsRequest) (*ListFrontendsResponse, error) {
	return &ListFrontendsResponse{FrontendNames: s.checker.ListFrontends()}, nil
}

// GetFrontend returns configuration details for a single frontend.
func (s *Server) GetFrontend(_ context.Context, req *GetFrontendRequest) (*FrontendInfo, error) {
	fe, ok := s.checker.GetFrontend(req.Name)
	if !ok {
		return nil, status.Errorf(codes.NotFound, "frontend %q not found", req.Name)
	}
	return frontendToProto(req.Name, fe, s.checker), nil
}

// ListBackends returns the names of all active backends.
func (s *Server) ListBackends(_ context.Context, _ *ListBackendsRequest) (*ListBackendsResponse, error) {
	return &ListBackendsResponse{BackendNames: s.checker.ListBackends()}, nil
}

// GetBackend returns health state for a backend by name.
func (s *Server) GetBackend(_ context.Context, req *GetBackendRequest) (*BackendInfo, error) {
	b, ok := s.checker.GetBackend(req.Name)
	if !ok {
		return nil, status.Errorf(codes.NotFound, "backend %q not found", req.Name)
	}
	return backendToProto(b), nil
}

// PauseBackend pauses health checking for a backend by name.
func (s *Server) PauseBackend(_ context.Context, req *BackendRequest) (*BackendInfo, error) {
	b, err := s.checker.PauseBackend(req.Name)
	if err != nil {
		return nil, status.Errorf(codes.FailedPrecondition, "%v", err)
	}
	return backendToProto(b), nil
}

// ResumeBackend resumes health checking for a backend by name.
func (s *Server) ResumeBackend(_ context.Context, req *BackendRequest) (*BackendInfo, error) {
	b, err := s.checker.ResumeBackend(req.Name)
	if err != nil {
		return nil, status.Errorf(codes.FailedPrecondition, "%v", err)
	}
	return backendToProto(b), nil
}

// EnableBackend re-enables a previously disabled backend.
func (s *Server) EnableBackend(_ context.Context, req *BackendRequest) (*BackendInfo, error) {
	b, ok := s.checker.EnableBackend(req.Name)
	if !ok {
		return nil, status.Errorf(codes.NotFound, "backend %q not found", req.Name)
	}
	return backendToProto(b), nil
}

// DisableBackend disables a backend, stopping its probe goroutine.
func (s *Server) DisableBackend(_ context.Context, req *BackendRequest) (*BackendInfo, error) {
	b, ok := s.checker.DisableBackend(req.Name)
	if !ok {
		return nil, status.Errorf(codes.NotFound, "backend %q not found", req.Name)
	}
	return backendToProto(b), nil
}

// SetFrontendPoolBackendWeight updates the weight of a backend in a pool
// and immediately pushes the change into VPP via a targeted single-VIP
// sync. When req.Flush is true the backend's AS row is rewritten with
// lb_as_set_weight(is_flush=true), which tears down VPP's flow table for
// that AS so existing sessions are dropped; when false the flow table is
// left alone and only Maglev's new-bucket mapping is updated, so existing
// sessions keep reaching this backend until they naturally drain.
func (s *Server) SetFrontendPoolBackendWeight(_ context.Context, req *SetWeightRequest) (*FrontendInfo, error) {
	if req.Weight < 0 || req.Weight > 100 {
		return nil, status.Errorf(codes.InvalidArgument, "weight %d out of range [0, 100]", req.Weight)
	}
	fe, err := s.checker.SetFrontendPoolBackendWeight(req.Frontend, req.Pool, req.Backend, int(req.Weight))
	if err != nil {
		return nil, status.Errorf(codes.NotFound, "%v", err)
	}

	// Push the change into VPP so the operator doesn't have to wait
	// for the periodic 30s reconcile to pick it up. Silently skipped
	// when VPP integration is disabled — the mutation still lands in
	// config and any future sync will reconcile it.
	if s.vppClient != nil && s.vppClient.IsConnected() {
		cfg := s.checker.Config()
		flushAddr := ""
		if req.Flush {
			if b, ok := cfg.Backends[req.Backend]; ok && b.Address != nil {
				flushAddr = b.Address.String()
			}
		}
		if err := s.vppClient.SyncLBStateVIP(cfg, req.Frontend, flushAddr); err != nil && !errors.Is(err, vpp.ErrFrontendNotFound) {
			slog.Warn("set-weight-sync",
				"frontend", req.Frontend, "backend", req.Backend,
				"weight", req.Weight, "flush", req.Flush, "err", err)
		}
	}

	return frontendToProto(req.Frontend, fe, s.checker), nil
}

// ListHealthChecks returns the names of all configured health checks.
func (s *Server) ListHealthChecks(_ context.Context, _ *ListHealthChecksRequest) (*ListHealthChecksResponse, error) {
	return &ListHealthChecksResponse{Names: s.checker.ListHealthChecks()}, nil
}

// GetHealthCheck returns the full configuration for a health check by name.
func (s *Server) GetHealthCheck(_ context.Context, req *GetHealthCheckRequest) (*HealthCheckInfo, error) {
	hc, ok := s.checker.GetHealthCheck(req.Name)
	if !ok {
		return nil, status.Errorf(codes.NotFound, "healthcheck %q not found", req.Name)
	}
	return healthCheckToProto(req.Name, hc), nil
}

// WatchEvents streams events to the client. On connect, the current state of
// every backend and/or frontend is sent as a synthetic event. Afterwards,
// live events are forwarded based on the filter flags in req. An unset (nil)
// flag defaults to true (subscribe). An empty log_level defaults to "info".
func (s *Server) WatchEvents(req *WatchRequest, stream Maglev_WatchEventsServer) error {
	wantLog := req.Log == nil || *req.Log
	wantBackend := req.Backend == nil || *req.Backend
	wantFrontend := req.Frontend == nil || *req.Frontend

	logLevel := slog.LevelInfo
	if req.LogLevel != "" {
		if err := logLevel.UnmarshalText([]byte(req.LogLevel)); err != nil {
			return status.Errorf(codes.InvalidArgument, "invalid log_level %q: must be debug, info, warn, or error", req.LogLevel)
		}
	}

	// Subscribe to log events (nil channel blocks forever when not wanted).
	var logCh <-chan *LogEvent
	if wantLog && s.logs != nil {
		var unsub func()
		logCh, unsub = s.logs.Subscribe(logLevel)
		defer unsub()
	}

	// Subscribe to the checker event stream once; we demultiplex backend
	// and frontend events in the select below. Skip the subscription if
	// neither kind is wanted.
	var eventCh <-chan checker.Event
	if wantBackend || wantFrontend {
		var unsub func()
		eventCh, unsub = s.checker.Subscribe()
		defer unsub()
	}

	// Send initial state snapshot: one synthetic event per existing backend
	// (if wanted), and one per existing frontend (if wanted). Clients that
	// connect mid-flight see the current state immediately instead of
	// waiting for the next transition.
	if wantBackend {
		for _, name := range s.checker.ListBackends() {
			snap, ok := s.checker.GetBackend(name)
			if !ok {
				continue
			}
			ev := &Event{Event: &Event_Backend{Backend: &BackendEvent{
				BackendName: name,
				Transition: &TransitionRecord{
					From:     snap.Health.State.String(),
					To:       snap.Health.State.String(),
					AtUnixNs: 0,
				},
			}}}
			if err := stream.Send(ev); err != nil {
				return err
			}
		}
	}
	if wantFrontend {
		for _, name := range s.checker.ListFrontends() {
			fs, ok := s.checker.FrontendState(name)
			if !ok {
				continue
			}
			ev := &Event{Event: &Event_Frontend{Frontend: &FrontendEvent{
				FrontendName: name,
				Transition: &TransitionRecord{
					From:     fs.String(),
					To:       fs.String(),
					AtUnixNs: 0,
				},
			}}}
			if err := stream.Send(ev); err != nil {
				return err
			}
		}
	}

	for {
		select {
		case <-s.ctx.Done():
			return status.Error(codes.Unavailable, "server shutting down")
		case <-stream.Context().Done():
			return nil
		case le, ok := <-logCh:
			if !ok {
				return nil
			}
			if err := stream.Send(&Event{Event: &Event_Log{Log: le}}); err != nil {
				return err
			}
		case e, ok := <-eventCh:
			if !ok {
				return nil
			}
			if e.FrontendTransition != nil {
				if !wantFrontend {
					continue
				}
				if err := stream.Send(&Event{Event: &Event_Frontend{Frontend: &FrontendEvent{
					FrontendName: e.FrontendName,
					Transition: &TransitionRecord{
						From:     e.FrontendTransition.From.String(),
						To:       e.FrontendTransition.To.String(),
						AtUnixNs: e.FrontendTransition.At.UnixNano(),
					},
				}}}); err != nil {
					return err
				}
				continue
			}
			if !wantBackend {
				continue
			}
			if err := stream.Send(&Event{Event: &Event_Backend{Backend: &BackendEvent{
				BackendName: e.BackendName,
				Transition:  transitionToProto(e.Transition),
			}}}); err != nil {
				return err
			}
		}
	}
}

// CheckConfig reads and validates the configuration file, returning a
// structured result that distinguishes YAML parse errors from semantic errors.
func (s *Server) CheckConfig(_ context.Context, _ *CheckConfigRequest) (*CheckConfigResponse, error) {
	slog.Info("config-check-start", "path", s.configPath)
	_, result := config.Check(s.configPath)
	resp := &CheckConfigResponse{
		Ok:            result.OK(),
		ParseError:    result.ParseError,
		SemanticError: result.SemanticError,
	}
	if result.OK() {
		slog.Info("config-check-done", "result", "ok")
	} else if result.ParseError != "" {
		slog.Info("config-check-done", "result", "failed", "type", "parse", "err", result.ParseError)
	} else {
		slog.Info("config-check-done", "result", "failed", "type", "semantic", "err", result.SemanticError)
	}
	return resp, nil
}

// ReloadConfig checks the configuration file and, if valid, applies it to the
// running checker. This is the same code path used by SIGHUP.
func (s *Server) ReloadConfig(_ context.Context, _ *ReloadConfigRequest) (*ReloadConfigResponse, error) {
	return s.doReloadConfig(), nil
}

// TriggerReload performs a config check and reload. Intended for use by the
// SIGHUP handler so that signals and gRPC share the same code path.
func (s *Server) TriggerReload() {
	s.doReloadConfig()
}

func (s *Server) doReloadConfig() *ReloadConfigResponse {
	slog.Info("config-reload-start")
	newCfg, result := config.Check(s.configPath)
	if !result.OK() {
		if result.ParseError != "" {
			slog.Error("config-check-failed", "type", "parse", "err", result.ParseError)
		} else {
			slog.Error("config-check-failed", "type", "semantic", "err", result.SemanticError)
		}
		return &ReloadConfigResponse{
			ParseError:    result.ParseError,
			SemanticError: result.SemanticError,
		}
	}
	if err := s.checker.Reload(s.ctx, newCfg); err != nil {
		slog.Error("checker-reload-error", "err", err)
		return &ReloadConfigResponse{
			ReloadError: err.Error(),
		}
	}
	// Push new global LB conf to VPP if anything changed. SetLBConf is a
	// no-op when VPP isn't connected or when the values are unchanged.
	if s.vppClient != nil {
		if err := s.vppClient.SetLBConf(newCfg); err != nil {
			slog.Warn("vpp-lb-conf-set-failed", "err", err)
		}
	}
	slog.Info("config-reload-done", "frontends", len(newCfg.Frontends))
	return &ReloadConfigResponse{Ok: true}
}

// GetVPPInfo returns VPP version and runtime information.
func (s *Server) GetVPPInfo(_ context.Context, _ *GetVPPInfoRequest) (*VPPInfo, error) {
	if s.vppClient == nil {
		return nil, status.Error(codes.Unavailable, "VPP integration is disabled")
	}
	info, err := s.vppClient.GetInfo()
	if err != nil {
		return nil, status.Errorf(codes.Unavailable, "%v", err)
	}
	var boottimeNs int64
	if !info.BootTime.IsZero() {
		boottimeNs = info.BootTime.UnixNano()
	}
	return &VPPInfo{
		Version:        info.Version,
		BuildDate:      info.BuildDate,
		BuildDirectory: info.BuildDirectory,
		Pid:            info.PID,
		BoottimeNs:     boottimeNs,
		ConnecttimeNs:  info.ConnectedSince.UnixNano(),
	}, nil
}

// GetVPPLBState returns a snapshot of the VPP load-balancer plugin state.
func (s *Server) GetVPPLBState(_ context.Context, _ *GetVPPLBStateRequest) (*VPPLBState, error) {
	if s.vppClient == nil {
		return nil, status.Error(codes.Unavailable, "VPP integration is disabled")
	}
	state, err := s.vppClient.GetLBStateAll()
	if err != nil {
		return nil, status.Errorf(codes.Unavailable, "%v", err)
	}
	return lbStateToProto(state), nil
}

// GetVPPLBCounters returns the most recent per-VIP counter snapshot
// captured by the client's 5s scrape loop. The call is served from an
// in-process cache and does not hit VPP. An empty response is returned
// when VPP is disconnected or no scrape has completed yet.
//
// Per-backend counters are deliberately not included: the LB plugin's
// forwarding node bypasses the ip{4,6}-lookup path, so the FIB's
// /net/route/to counter at a backend's stats_index never ticks for
// LB-forwarded traffic (see internal/vpp/lbstats.go::scrapeLBStats
// for the full chain). Exposing a zero-valued "packets" column would
// just mislead operators.
func (s *Server) GetVPPLBCounters(_ context.Context, _ *GetVPPLBCountersRequest) (*VPPLBCounters, error) {
	if s.vppClient == nil {
		return nil, status.Error(codes.Unavailable, "VPP integration is disabled")
	}
	out := &VPPLBCounters{}
	for _, v := range s.vppClient.VIPStats() {
		out.Vips = append(out.Vips, &VPPLBVIPCounters{
			Prefix:          v.Prefix,
			Protocol:        v.Protocol,
			Port:            uint32(v.Port),
			NextPacket:      v.NextPkt,
			FirstPacket:     v.FirstPkt,
			UntrackedPacket: v.Untracked,
			NoServer:        v.NoServer,
			Packets:         v.Packets,
			Bytes:           v.Bytes,
		})
	}
	sort.Slice(out.Vips, func(i, j int) bool {
		if out.Vips[i].Prefix != out.Vips[j].Prefix {
			return out.Vips[i].Prefix < out.Vips[j].Prefix
		}
		if out.Vips[i].Protocol != out.Vips[j].Protocol {
			return out.Vips[i].Protocol < out.Vips[j].Protocol
		}
		return out.Vips[i].Port < out.Vips[j].Port
	})
	return out, nil
}

// SyncVPPLBState runs the LB reconciler. With frontend_name unset it does a
// full sync (SyncLBStateAll), which may remove stale VIPs. With frontend_name
// set it does a single-VIP sync (SyncLBStateVIP) that only adds/updates.
func (s *Server) SyncVPPLBState(_ context.Context, req *SyncVPPLBStateRequest) (*SyncVPPLBStateResponse, error) {
	if s.vppClient == nil {
		return nil, status.Error(codes.Unavailable, "VPP integration is disabled")
	}
	cfg := s.checker.Config()
	if req.FrontendName != nil && *req.FrontendName != "" {
		if err := s.vppClient.SyncLBStateVIP(cfg, *req.FrontendName, ""); err != nil {
			if errors.Is(err, vpp.ErrFrontendNotFound) {
				return nil, status.Errorf(codes.NotFound, "%v", err)
			}
			return nil, status.Errorf(codes.Unavailable, "%v", err)
		}
		return &SyncVPPLBStateResponse{}, nil
	}
	if err := s.vppClient.SyncLBStateAll(cfg); err != nil {
		return nil, status.Errorf(codes.Unavailable, "%v", err)
	}
	return &SyncVPPLBStateResponse{}, nil
}

// lbStateToProto converts the vpp package's LBState into the proto message.
func lbStateToProto(s *vpp.LBState) *VPPLBState {
	out := &VPPLBState{
		Conf: &VPPLBConf{
			Ip4SrcAddress:        ipStringOrEmpty(s.Conf.IP4SrcAddress),
			Ip6SrcAddress:        ipStringOrEmpty(s.Conf.IP6SrcAddress),
			StickyBucketsPerCore: s.Conf.StickyBucketsPerCore,
			FlowTimeout:          s.Conf.FlowTimeout,
		},
	}
	for _, v := range s.VIPs {
		pv := &VPPLBVIP{
			Prefix:          v.Prefix.String(),
			Protocol:        uint32(v.Protocol),
			Port:            uint32(v.Port),
			Encap:           v.Encap,
			FlowTableLength: uint32(v.FlowTableLength),
			SrcIpSticky:     v.SrcIPSticky,
		}
		for _, a := range v.ASes {
			var ts int64
			if !a.InUseSince.IsZero() {
				ts = a.InUseSince.UnixNano()
			}
			pv.ApplicationServers = append(pv.ApplicationServers, &VPPLBAS{
				Address:      a.Address.String(),
				Weight:       uint32(a.Weight),
				Flags:        uint32(a.Flags),
				NumBuckets:   a.NumBuckets,
				InUseSinceNs: ts,
			})
		}
		out.Vips = append(out.Vips, pv)
	}
	return out
}

func ipStringOrEmpty(ip net.IP) string {
	if len(ip) == 0 || ip.IsUnspecified() {
		return ""
	}
	return ip.String()
}

// ---- conversion helpers ----------------------------------------------------

func frontendToProto(name string, fe config.Frontend, src vpp.StateSource) *FrontendInfo {
	// Compute the state-aware effective weights once; these reflect the
	// pool-failover logic and what would be programmed into VPP.
	eff := vpp.EffectiveWeights(fe, src)
	pools := make([]*PoolInfo, 0, len(fe.Pools))
	for poolIdx, p := range fe.Pools {
		pi := &PoolInfo{Name: p.Name}
		for bName, pb := range p.Backends {
			pi.Backends = append(pi.Backends, &PoolBackendInfo{
				Name:            bName,
				Weight:          int32(pb.Weight),
				EffectiveWeight: int32(eff[poolIdx][bName]),
			})
		}
		pools = append(pools, pi)
	}
	return &FrontendInfo{
		Name:        name,
		Address:     fe.Address.String(),
		Protocol:    fe.Protocol,
		Port:        uint32(fe.Port),
		Description: fe.Description,
		Pools:       pools,
		SrcIpSticky: fe.SrcIPSticky,
	}
}

func backendToProto(snap checker.BackendSnapshot) *BackendInfo {
	info := &BackendInfo{
		Name:        snap.Health.Name,
		Address:     snap.Health.Address.String(),
		State:       snap.Health.State.String(),
		Enabled:     snap.Config.Enabled,
		Healthcheck: snap.Config.HealthCheck,
	}
	for _, t := range snap.Health.Transitions {
		info.Transitions = append(info.Transitions, transitionToProto(t))
	}
	return info
}

func healthCheckToProto(name string, hc config.HealthCheck) *HealthCheckInfo {
	info := &HealthCheckInfo{
		Name:           name,
		Type:           hc.Type,
		Port:           uint32(hc.Port),
		IntervalNs:     hc.Interval.Nanoseconds(),
		FastIntervalNs: hc.FastInterval.Nanoseconds(),
		DownIntervalNs: hc.DownInterval.Nanoseconds(),
		TimeoutNs:      hc.Timeout.Nanoseconds(),
		Rise:           int32(hc.Rise),
		Fall:           int32(hc.Fall),
	}
	if hc.ProbeIPv4Src != nil {
		info.ProbeIpv4Src = hc.ProbeIPv4Src.String()
	}
	if hc.ProbeIPv6Src != nil {
		info.ProbeIpv6Src = hc.ProbeIPv6Src.String()
	}
	if hc.HTTP != nil {
		re := ""
		if hc.HTTP.ResponseRegexp != nil {
			re = hc.HTTP.ResponseRegexp.String()
		}
		info.Http = &HTTPCheckParams{
			Path:               hc.HTTP.Path,
			Host:               hc.HTTP.Host,
			ResponseCodeMin:    int32(hc.HTTP.ResponseCodeMin),
			ResponseCodeMax:    int32(hc.HTTP.ResponseCodeMax),
			ResponseRegexp:     re,
			ServerName:         hc.HTTP.ServerName,
			InsecureSkipVerify: hc.HTTP.InsecureSkipVerify,
		}
	}
	if hc.TCP != nil {
		info.Tcp = &TCPCheckParams{
			Ssl:                hc.TCP.SSL,
			ServerName:         hc.TCP.ServerName,
			InsecureSkipVerify: hc.TCP.InsecureSkipVerify,
		}
	}
	return info
}

func transitionToProto(t health.Transition) *TransitionRecord {
	return &TransitionRecord{
		From:     t.From.String(),
		To:       t.To.String(),
		AtUnixNs: t.At.UnixNano(),
	}
}

// Ensure net.IP is imported (used via b.Address.String()).
var _ = net.IP{}