Files
restic-manager/internal/server/ws/hub.go
T
steve f34773b505 phase 1: WS transport, enrollment, agent that hellos and heartbeats
Lands the protocol layer end-to-end: an agent can be enrolled
through the operator UI, store credentials, dial back to the server
over WS, complete the protocol_version handshake, and stay
connected with periodic heartbeats.

Server side:
- P1-09 ws.Hub: one Conn per host_id, last-write-wins eviction,
  json envelope writer with a write mutex, reader, error envelopes.
- P1-09 ws.AgentHandler: bearer-auth, accept upgrade, hello-stage
  (10s deadline, protocol_version checked against
  api.MinAgentProtocolVersion → ErrProtocolTooOld with help URL on
  reject), main read loop, defer hub register/unregister.
- P1-10 POST /api/agents/enroll consumes a one-time token, mints a
  persistent agent bearer (sha-256 stored), creates a host row.
- P1-10 POST /api/enrollment-tokens (operator, session-auth)
  issues a 1h one-time token.
- P1-11 hello upserts agent_version + restic_version +
  protocol_version on the host row, flips status to online.
- P1-12 heartbeat touches last_seen_at; background sweeper marks
  hosts offline after 90s without one.
- store: hosts table accessors, host_schedule_version,
  enrollment_tokens FK on consumed_host dropped (audit-only field;
  the token gets burned before the host row exists).

Agent side:
- P1-13 internal/agent/config: yaml at /etc/restic-manager/agent.yaml,
  atomic Save (tmp+fsync+rename), Enrolled() helper.
- P1-15 internal/agent/wsclient: dial with bearer + optional
  TLS cert pinning (sha-256 of leaf), exponential backoff with
  jitter (1s → 60s cap), heartbeat goroutine, fatal handling for
  ErrProtocolTooOld.
- P1-15 wsclient.Enroll: HTTP POST /api/agents/enroll with sysinfo.
- P1-17 internal/agent/sysinfo: hostname/OS/arch/restic-version
  collection. restic detected by `restic version` parse; absent
  restic doesn't block startup.
- cmd/agent: -enroll-server / -enroll-token flags drive first-run
  enrollment then exit (so the install script can hand off to
  systemd to run the persistent service).

End-to-end smoke verified: bootstrap → login → issue token →
enroll → run agent → server logs `ws agent connected` with the
right host_id and protocol_version 1.

All tests still pass.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-01 00:39:00 +01:00

146 lines
4.1 KiB
Go

// Package ws hosts the WebSocket transport for agent ↔ server. The
// Hub tracks one active connection per host id; subsequent connections
// from the same host evict the prior one (last-write-wins).
package ws
import (
"context"
"encoding/json"
"errors"
"fmt"
"log/slog"
"sync"
"time"
"github.com/coder/websocket"
"gitea.dcglab.co.uk/steve/restic-manager/internal/api"
)
// Hub owns the live agent connections and routes messages.
type Hub struct {
mu sync.RWMutex
conns map[string]*Conn // hostID → conn
}
// NewHub returns an empty hub.
func NewHub() *Hub {
return &Hub{conns: make(map[string]*Conn)}
}
// Conn is one agent WS connection. Send is safe for concurrent use;
// Read is single-reader (the connection's run loop).
type Conn struct {
HostID string
c *websocket.Conn
writeMu sync.Mutex
}
// Register installs c as the canonical connection for hostID. Any
// previous connection for that host is closed.
func (h *Hub) Register(hostID string, c *Conn) {
h.mu.Lock()
if prev, ok := h.conns[hostID]; ok {
// Best-effort close — a stuck old socket shouldn't block new one.
go func(old *Conn) {
_ = old.c.Close(websocket.StatusPolicyViolation, "superseded")
}(prev)
}
h.conns[hostID] = c
h.mu.Unlock()
}
// Unregister removes c iff it is still the canonical conn (a race
// where a newer conn already replaced it must not unregister it).
func (h *Hub) Unregister(hostID string, c *Conn) {
h.mu.Lock()
if cur, ok := h.conns[hostID]; ok && cur == c {
delete(h.conns, hostID)
}
h.mu.Unlock()
}
// Send delivers an envelope to the host if connected. Returns an error
// if the host is offline; caller may queue the message for later.
func (h *Hub) Send(ctx context.Context, hostID string, env api.Envelope) error {
h.mu.RLock()
c, ok := h.conns[hostID]
h.mu.RUnlock()
if !ok {
return fmt.Errorf("ws: host %q is offline", hostID)
}
return c.Send(ctx, env)
}
// Connected reports whether hostID has an active connection.
func (h *Hub) Connected(hostID string) bool {
h.mu.RLock()
_, ok := h.conns[hostID]
h.mu.RUnlock()
return ok
}
// ----- Conn methods --------------------------------------------------
// NewConn wraps a freshly-accepted websocket for a given hostID.
func NewConn(hostID string, c *websocket.Conn) *Conn {
return &Conn{HostID: hostID, c: c}
}
// Send writes an envelope as a JSON text message. Concurrent calls
// are serialised; the underlying socket is not safe for parallel
// writers.
func (c *Conn) Send(ctx context.Context, env api.Envelope) error {
c.writeMu.Lock()
defer c.writeMu.Unlock()
raw, err := json.Marshal(env)
if err != nil {
return fmt.Errorf("ws: marshal envelope: %w", err)
}
return c.c.Write(ctx, websocket.MessageText, raw)
}
// SendError writes an error envelope and closes the socket. Used by
// the hello handshake when an agent is rejected.
func (c *Conn) SendError(ctx context.Context, code api.ErrorCode, msg, helpURL string) {
env, err := api.Marshal(api.MsgError, "", api.ErrorPayload{
Code: code, Message: msg, HelpURL: helpURL,
})
if err == nil {
writeCtx, cancel := context.WithTimeout(ctx, 2*time.Second)
defer cancel()
_ = c.Send(writeCtx, env)
}
_ = c.c.Close(websocket.StatusPolicyViolation, string(code))
}
// Close shuts the socket down with a normal-closure status code.
func (c *Conn) Close() error {
return c.c.Close(websocket.StatusNormalClosure, "")
}
// Read pulls the next JSON envelope off the wire. The caller's
// context controls cancellation and timeouts (e.g. read deadlines).
func (c *Conn) Read(ctx context.Context) (api.Envelope, error) {
mt, raw, err := c.c.Read(ctx)
if err != nil {
return api.Envelope{}, err
}
if mt != websocket.MessageText {
return api.Envelope{}, errors.New("ws: expected text frame")
}
var env api.Envelope
if err := json.Unmarshal(raw, &env); err != nil {
return api.Envelope{}, fmt.Errorf("ws: unmarshal envelope: %w", err)
}
return env, nil
}
// ----- helpers -------------------------------------------------------
// LogValue emits a slog-friendly representation of a Conn.
func (c *Conn) LogValue() slog.Value {
return slog.GroupValue(slog.String("host_id", c.HostID))
}