opensource/coder
3
0
Fork 0
mirror of https://github.com/coder/coder.git synced 2026-09-24 15:04:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat: tag chat-originating agent logs with chat_id (#25019)

Browse Source 
Workspace-agent logs emitted while serving chatd-driven requests were
not correlated with the originating chat, making agent logs hard to
attribute to the corresponding/originating chat.

This adds agent-side chat context middleware that parses `Coder-Chat-Id`
once, enriches agent access logs and structured handler/background logs,
and adds a chatd bridge log when chat headers are attached to an agent
connection.

Closes CODAGT-324
This commit is contained in:
Ethan
2026-05-08 13:25:30 +10:00
committed by GitHub
parent e9f0385198
commit 3a9080fff6
15 changed files with 380 additions and 95 deletions
Download Patch File Download Diff File Expand all files Collapse all files
agent/x/agentdesktop/api.go
+17 -13
View File
@@ -16,6 +16,7 @@ import (
"github.com/google/uuid"
"cdr.dev/slog/v3"
"github.com/coder/coder/v2/agent/agentchat"
"github.com/coder/coder/v2/agent/agentssh"
"github.com/coder/coder/v2/coderd/httpapi"
"github.com/coder/coder/v2/codersdk"
@@ -85,6 +86,7 @@ func (a *API) Routes() http.Handler {
func (a *API) handleDesktopVNC(rw http.ResponseWriter, r *http.Request) {
ctx := r.Context()
logger := a.logger.With(agentchat.Fields(ctx)...)
// Start the desktop session (idempotent).
_, err := a.desktop.Start(ctx)
@@ -112,7 +114,7 @@ func (a *API) handleDesktopVNC(rw http.ResponseWriter, r *http.Request) {
CompressionMode: websocket.CompressionDisabled,
})
if err != nil {
a.logger.Error(ctx, "failed to accept websocket", slog.Error(err))
logger.Error(ctx, "failed to accept websocket", slog.Error(err))
return
}
@@ -128,6 +130,7 @@ func (a *API) handleDesktopVNC(rw http.ResponseWriter, r *http.Request) {
func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
ctx := r.Context()
logger := a.logger.With(agentchat.Fields(ctx)...)
handlerStart := a.clock.Now()
// Update last desktop action timestamp for idle recording monitor.
@@ -136,7 +139,7 @@ func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
// Ensure the desktop is running and grab native dimensions.
cfg, err := a.desktop.Start(ctx)
if err != nil {
a.logger.Warn(ctx, "handleAction: desktop.Start failed",
logger.Warn(ctx, "handleAction: desktop.Start failed",
slog.Error(err),
slog.F("elapsed_ms", a.clock.Since(handlerStart).Milliseconds()),
)
@@ -156,7 +159,7 @@ func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
return
}
a.logger.Info(ctx, "handleAction: started",
logger.Info(ctx, "handleAction: started",
slog.F("action", action.Action),
slog.F("elapsed_ms", a.clock.Since(handlerStart).Milliseconds()),
)
@@ -272,7 +275,7 @@ func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
x, y = scaleXY(x, y)
stepStart := a.clock.Now()
if err := a.desktop.Click(ctx, x, y, MouseButtonLeft); err != nil {
a.logger.Warn(ctx, "handleAction: Click failed",
logger.Warn(ctx, "handleAction: Click failed",
slog.F("action", "left_click"),
slog.F("step", "click"),
slog.F("step_ms", time.Since(stepStart).Milliseconds()),
@@ -285,7 +288,7 @@ func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
})
return
}
a.logger.Debug(ctx, "handleAction: Click completed",
logger.Debug(ctx, "handleAction: Click completed",
slog.F("action", "left_click"),
slog.F("step_ms", time.Since(stepStart).Milliseconds()),
slog.F("elapsed_ms", a.clock.Since(handlerStart).Milliseconds()),
@@ -473,7 +476,7 @@ func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
case <-ctx.Done():
// Context canceled; release the key immediately.
if err := a.desktop.KeyUp(ctx, *action.Text); err != nil {
a.logger.Warn(ctx, "handleAction: KeyUp after context cancel", slog.Error(err))
logger.Warn(ctx, "handleAction: KeyUp after context cancel", slog.Error(err))
}
return
case <-timer.C:
@@ -513,14 +516,14 @@ func (a *API) handleAction(rw http.ResponseWriter, r *http.Request) {
elapsedMs := a.clock.Since(handlerStart).Milliseconds()
if ctx.Err() != nil {
a.logger.Error(ctx, "handleAction: context canceled before writing response",
logger.Error(ctx, "handleAction: context canceled before writing response",
slog.F("action", action.Action),
slog.F("elapsed_ms", elapsedMs),
slog.Error(ctx.Err()),
)
return
}
a.logger.Info(ctx, "handleAction: writing response",
logger.Info(ctx, "handleAction: writing response",
slog.F("action", action.Action),
slog.F("elapsed_ms", elapsedMs),
)
@@ -609,6 +612,7 @@ func (a *API) handleRecordingStart(rw http.ResponseWriter, r *http.Request) {
func (a *API) handleRecordingStop(rw http.ResponseWriter, r *http.Request) {
ctx := r.Context()
logger := a.logger.With(agentchat.Fields(ctx)...)
recordingID, ok := a.decodeRecordingRequest(rw, r)
if !ok {
@@ -661,7 +665,7 @@ func (a *API) handleRecordingStop(rw http.ResponseWriter, r *http.Request) {
}()
if artifact.Size > workspacesdk.MaxRecordingSize {
a.logger.Warn(ctx, "recording file exceeds maximum size",
logger.Warn(ctx, "recording file exceeds maximum size",
slog.F("recording_id", recordingID),
slog.F("size", artifact.Size),
slog.F("max_size", workspacesdk.MaxRecordingSize),
@@ -677,7 +681,7 @@ func (a *API) handleRecordingStop(rw http.ResponseWriter, r *http.Request) {
// rejecting it here avoids streaming a large thumbnail over
// the wire for nothing.
if artifact.ThumbnailReader != nil && artifact.ThumbnailSize > workspacesdk.MaxThumbnailSize {
a.logger.Warn(ctx, "thumbnail file exceeds maximum size, omitting",
logger.Warn(ctx, "thumbnail file exceeds maximum size, omitting",
slog.F("recording_id", recordingID),
slog.F("size", artifact.ThumbnailSize),
slog.F("max_size", workspacesdk.MaxThumbnailSize),
@@ -701,13 +705,13 @@ func (a *API) handleRecordingStop(rw http.ResponseWriter, r *http.Request) {
"Content-Type": {"video/mp4"},
})
if err != nil {
a.logger.Warn(ctx, "failed to create video multipart part",
logger.Warn(ctx, "failed to create video multipart part",
slog.F("recording_id", recordingID),
slog.Error(err))
return
}
if _, err := io.Copy(videoPart, artifact.Reader); err != nil {
a.logger.Warn(ctx, "failed to write video multipart part",
logger.Warn(ctx, "failed to write video multipart part",
slog.F("recording_id", recordingID),
slog.Error(err))
return
@@ -719,7 +723,7 @@ func (a *API) handleRecordingStop(rw http.ResponseWriter, r *http.Request) {
"Content-Type": {"image/jpeg"},
})
if err != nil {
a.logger.Warn(ctx, "failed to create thumbnail multipart part",
logger.Warn(ctx, "failed to create thumbnail multipart part",
slog.F("recording_id", recordingID),
slog.Error(err))
return
agent/x/agentmcp/api.go
+6 -4
View File
@@ -8,6 +8,7 @@ import (
"github.com/go-chi/chi/v5"
"cdr.dev/slog/v3"
"github.com/coder/coder/v2/agent/agentchat"
"github.com/coder/coder/v2/coderd/httpapi"
"github.com/coder/coder/v2/codersdk"
"github.com/coder/coder/v2/codersdk/workspacesdk"
@@ -52,6 +53,7 @@ func (api *API) Routes() http.Handler {
// independent of config changes.
func (api *API) handleListTools(rw http.ResponseWriter, r *http.Request) {
ctx := r.Context()
logger := api.logger.With(agentchat.Fields(ctx)...)
// Check config freshness and reload if changed.
var reloaded bool
@@ -61,11 +63,11 @@ func (api *API) handleListTools(rw http.ResponseWriter, r *http.Request) {
// Categorize the error for operator debugging.
switch {
case errors.Is(err, context.Canceled):
api.logger.Warn(ctx, "mcp reload canceled by caller", slog.Error(err))
logger.Warn(ctx, "mcp reload canceled by caller", slog.Error(err))
case errors.Is(err, context.DeadlineExceeded):
api.logger.Warn(ctx, "mcp reload timed out", slog.Error(err))
logger.Warn(ctx, "mcp reload timed out", slog.Error(err))
default:
api.logger.Warn(ctx, "mcp reload failed", slog.Error(err))
logger.Warn(ctx, "mcp reload failed", slog.Error(err))
}
// Fall through to return whatever tools we have.
} else {
@@ -78,7 +80,7 @@ func (api *API) handleListTools(rw http.ResponseWriter, r *http.Request) {
// refreshes tools as part of the reload.
if r.URL.Query().Get("refresh") == "true" && !reloaded {
if err := api.manager.RefreshTools(ctx); err != nil {
api.logger.Warn(ctx, "failed to refresh MCP tools", slog.Error(err))
logger.Warn(ctx, "failed to refresh MCP tools", slog.Error(err))
}
}
agent/x/agentmcp/manager.go
+15 -5
View File
@@ -22,6 +22,7 @@ import (
tailscalesingleflight "tailscale.com/util/singleflight"
"cdr.dev/slog/v3"
"github.com/coder/coder/v2/agent/agentchat"
"github.com/coder/coder/v2/agent/agentexec"
"github.com/coder/coder/v2/agent/usershell"
"github.com/coder/coder/v2/buildinfo"
@@ -248,7 +249,8 @@ func (m *Manager) doReload(ctx context.Context, mcpConfigFiles []string) error {
// Refresh tools outside the lock to avoid blocking
// concurrent reads during network I/O.
if err := m.RefreshTools(ctx); err != nil {
m.logger.Warn(ctx, "failed to refresh MCP tools after connect", slog.Error(err))
logger := m.logger.With(agentchat.Fields(ctx)...)
logger.Warn(ctx, "failed to refresh MCP tools after connect", slog.Error(err))
}
return nil
}
@@ -257,6 +259,8 @@ func (m *Manager) doReload(ctx context.Context, mcpConfigFiles []string) error {
// list of server configs. Missing files are silently skipped;
// parse errors are logged and skipped.
func (m *Manager) parseAndDedup(ctx context.Context, mcpConfigFiles []string) ([]ServerConfig, map[string]fileSnapshot) {
logger := m.logger.With(agentchat.Fields(ctx)...)
// Stat before reading so the snapshot is conservatively old.
// If a file changes between stat and read, the snapshot
// records the old mtime, SnapshotChanged detects a mismatch
@@ -272,7 +276,7 @@ func (m *Manager) parseAndDedup(ctx context.Context, mcpConfigFiles []string) ([
if errors.Is(err, fs.ErrNotExist) {
continue
}
m.logger.Warn(ctx, "failed to parse MCP config",
logger.Warn(ctx, "failed to parse MCP config",
slog.F("path", configPath),
slog.Error(err),
)
@@ -334,6 +338,8 @@ func (m *Manager) classifyServers(wanted map[string]ServerConfig) (*serverDiff,
// connectAll runs connectServer in parallel for the given configs.
// Failed connects are logged and skipped.
func (m *Manager) connectAll(ctx context.Context, toConnect []ServerConfig) []connectedServer {
logger := m.logger.With(agentchat.Fields(ctx)...)
var (
mu sync.Mutex
connected []connectedServer
@@ -343,7 +349,7 @@ func (m *Manager) connectAll(ctx context.Context, toConnect []ServerConfig) []co
eg.Go(func() error {
c, err := m.connectServer(ctx, cfg)
if err != nil {
m.logger.Warn(ctx, "skipping MCP server",
logger.Warn(ctx, "skipping MCP server",
slog.F("server", cfg.Name),
slog.F("transport", cfg.Transport),
slog.Error(err),
@@ -481,6 +487,8 @@ func (m *Manager) CallTool(ctx context.Context, req workspacesdk.CallMCPToolRequ
// existing cached tools for servers that failed, so a single
// dead server doesn't block updates from healthy ones.
func (m *Manager) RefreshTools(ctx context.Context) error {
logger := m.logger.With(agentchat.Fields(ctx)...)
// Snapshot servers under read lock.
m.mu.RLock()
servers := make(map[string]*serverEntry, len(m.servers))
@@ -508,7 +516,7 @@ func (m *Manager) RefreshTools(ctx context.Context) error {
result, err := entry.client.ListTools(listCtx, mcp.ListToolsRequest{})
cancel()
if err != nil {
m.logger.Warn(ctx, "failed to list tools from MCP server",
logger.Warn(ctx, "failed to list tools from MCP server",
slog.F("server", name),
slog.Error(err),
)
@@ -670,12 +678,14 @@ func (m *Manager) createTransport(ctx context.Context, cfg ServerConfig) (transp
// updateEnv callback, then merges explicit overrides from the
// server config on top.
func (m *Manager) buildEnv(ctx context.Context, explicit map[string]string) []string {
logger := m.logger.With(agentchat.Fields(ctx)...)
env := usershell.SystemEnvInfo{}.Environ()
if m.updateEnv != nil {
var err error
env, err = m.updateEnv(env)
if err != nil {
m.logger.Warn(ctx, "failed to enrich MCP server environment",
logger.Warn(ctx, "failed to enrich MCP server environment",
slog.Error(err),
)
env = usershell.SystemEnvInfo{}.Environ()