fix: agentcontainers: fix flake when ctx cancelled while running docker inspect (#18526)

johnstcn · web-flow · commit 40667855b1cf · 2025-06-24T13:30:09.000+01:00
Should fix coder/internal#738
diff --git a/agent/agentcontainers/api.go b/agent/agentcontainers/api.go
@@ -378,7 +378,11 @@ func (api *API) updaterLoop() {
 	// and anyone looking to interact with the API.
 	api.logger.Debug(api.ctx, "performing initial containers update")
 	if err := api.updateContainers(api.ctx); err != nil {
-		api.logger.Error(api.ctx, "initial containers update failed", slog.Error(err))
+		if errors.Is(err, context.Canceled) {
+			api.logger.Warn(api.ctx, "initial containers update canceled", slog.Error(err))
+		} else {
+			api.logger.Error(api.ctx, "initial containers update failed", slog.Error(err))
+		}
 	} else {
 		api.logger.Debug(api.ctx, "initial containers update complete")
 	}
@@ -399,7 +403,11 @@ func (api *API) updaterLoop() {
 		case api.updateTrigger <- done:
 			err := <-done
 			if err != nil {
-				api.logger.Error(api.ctx, "updater loop ticker failed", slog.Error(err))
+				if errors.Is(err, context.Canceled) {
+					api.logger.Warn(api.ctx, "updater loop ticker canceled", slog.Error(err))
+				} else {
+					api.logger.Error(api.ctx, "updater loop ticker failed", slog.Error(err))
+				}
 			}
 		default:
 			api.logger.Debug(api.ctx, "updater loop ticker skipped, update in progress")
diff --git a/agent/agentcontainers/containers_dockercli.go b/agent/agentcontainers/containers_dockercli.go
@@ -311,6 +311,10 @@ func (dcli *dockerCLI) List(ctx context.Context) (codersdk.WorkspaceAgentListCon
 // container IDs and returns the parsed output.
 // The stderr output is also returned for logging purposes.
 func runDockerInspect(ctx context.Context, execer agentexec.Execer, ids ...string) (stdout, stderr []byte, err error) {
+	if ctx.Err() != nil {
+		// If the context is done, we don't want to run the command.
+		return []byte{}, []byte{}, ctx.Err()
+	}
 	var stdoutBuf, stderrBuf bytes.Buffer
 	cmd := execer.CommandContext(ctx, "docker", append([]string{"inspect"}, ids...)...)
 	cmd.Stdout = &stdoutBuf
@@ -319,6 +323,12 @@ func runDockerInspect(ctx context.Context, execer agentexec.Execer, ids ...strin
 	stdout = bytes.TrimSpace(stdoutBuf.Bytes())
 	stderr = bytes.TrimSpace(stderrBuf.Bytes())
 	if err != nil {
+		if ctx.Err() != nil {
+			// If the context was canceled while running the command,
+			// return the context error instead of the command error,
+			// which is likely to be "signal: killed".
+			return stdout, stderr, ctx.Err()
+		}
 		if bytes.Contains(stderr, []byte("No such object:")) {
 			// This can happen if a container is deleted between the time we check for its existence and the time we inspect it.
 			return stdout, stderr, nil
