fix(coderd): render collected_at as UTC RFC3339 in the agent metadata aggregate (#27991)

Follow-up to #27934; this fix was pushed to the branch after the
squash-merge and missed it.

`jsonb_build_object` renders timestamptz in the session `TimeZone`,
which Coder never pins, and `collected_at` defaults to year 1 until the
agent's first report. On a non-UTC Postgres session a
registered-but-never-collected item renders with an LMT second-offset
(even `BC`, e.g. `0001-12-31T19:03:58-04:56:02 BC`), which Go's RFC3339
parsing rejects - a 500 for the entire list page whenever
`include_agent_metadata` is used.

- `to_char(... AT TIME ZONE 'UTC', ...)` pins the rendering;
never-collected items round-trip as Go's zero time.
- The test now runs against a named-zone database
(`dbtestutil.WithTimezone("America/Caracas")`) and requests a registered
but never-collected key; it reproduces the 500 without the fix.

Also contains the failure mode Go-side: an unparsable aggregate now
degrades to missing metadata for that workspace (with a warning log)
instead of failing the entire page. The SQL fix prevents the known
cause; the containment covers any future one. The test still catches
regressions because it asserts the metadata values, not just a 200.

---

Authored by Coder Agents on behalf of @Emyrk.

coderd/database/queries.sql.go

@@ -38914,7 +38914,11 @@ SELECT
 					'error', workspace_agent_metadata.error,
 					'timeout', workspace_agent_metadata.timeout,
 					'interval', workspace_agent_metadata.interval,
-					'collected_at', workspace_agent_metadata.collected_at,
+					-- Rendered explicitly as UTC RFC3339: jsonb renders
+					-- timestamptz in the session TimeZone, and the year-1
+					-- default of collected_at renders named zones with
+					-- LMT second-offsets (even BC), which Go rejects.
+					'collected_at', to_char(workspace_agent_metadata.collected_at AT TIME ZONE 'UTC', 'YYYY-MM-DD"T"HH24:MI:SS.US"Z"'),
 					'display_order', workspace_agent_metadata.display_order
 				))
 			FROM

coderd/database/queries/workspaces.sql

@@ -494,7 +494,11 @@ SELECT
 					'error', workspace_agent_metadata.error,
 					'timeout', workspace_agent_metadata.timeout,
 					'interval', workspace_agent_metadata.interval,
-					'collected_at', workspace_agent_metadata.collected_at,
+					-- Rendered explicitly as UTC RFC3339: jsonb renders
+					-- timestamptz in the session TimeZone, and the year-1
+					-- default of collected_at renders named zones with
+					-- LMT second-offsets (even BC), which Go rejects.
+					'collected_at', to_char(workspace_agent_metadata.collected_at AT TIME ZONE 'UTC', 'YYYY-MM-DD"T"HH24:MI:SS.US"Z"'),
 					'display_order', workspace_agent_metadata.display_order
 				))
 			FROM

coderd/workspaces.go

@@ -249,14 +249,7 @@ func (api *API) workspaces(rw http.ResponseWriter, r *http.Request) {
 	}
 
 	if len(filter.IncludeAgentMetadata) > 0 {
-		err = attachAgentMetadata(wss, workspaceRows)
-		if err != nil {
-			httpapi.Write(ctx, rw, http.StatusInternalServerError, codersdk.Response{
-				Message: "Internal error converting agent metadata.",
-				Detail:  err.Error(),
-			})
-			return
-		}
+		attachAgentMetadata(ctx, api.Logger, wss, workspaceRows)
 	}
 
 	httpapi.Write(ctx, rw, http.StatusOK, codersdk.WorkspacesResponse{
@@ -2771,8 +2764,11 @@ func (api *API) workspaceData(ctx context.Context, workspaces []database.Workspa
 
 // attachAgentMetadata maps the agent metadata the workspaces query
 // aggregated per workspace onto the agents in the converted response.
-// Each aggregated datum carries its workspace_agent_id.
-func attachAgentMetadata(workspaces []codersdk.Workspace, rows []database.GetWorkspacesRow) error {
+// Each aggregated datum carries its workspace_agent_id. An unparsable
+// aggregate degrades to missing metadata for that workspace rather
+// than failing the page: the expansion is best-effort decoration on
+// top of the list.
+func attachAgentMetadata(ctx context.Context, logger slog.Logger, workspaces []codersdk.Workspace, rows []database.GetWorkspacesRow) {
 	byAgent := map[uuid.UUID][]database.WorkspaceAgentMetadatum{}
 	for _, row := range rows {
 		if len(row.AgentMetadata) == 0 {
@@ -2781,7 +2777,11 @@ func attachAgentMetadata(workspaces []codersdk.Workspace, rows []database.GetWor
 		var metadata database.AgentMetadataAggregate
 		err := metadata.Scan(row.AgentMetadata)
 		if err != nil {
-			return xerrors.Errorf("scan agent metadata for workspace %q: %w", row.ID, err)
+			logger.Warn(ctx, "scan agent metadata, omitting it for the workspace",
+				slog.F("workspace_id", row.ID),
+				slog.Error(err),
+			)
+			continue
 		}
 		for _, datum := range metadata {
 			byAgent[datum.WorkspaceAgentID] = append(byAgent[datum.WorkspaceAgentID], datum)
@@ -2798,7 +2798,6 @@ func attachAgentMetadata(workspaces []codersdk.Workspace, rows []database.GetWor
 			}
 		}
 	}
-	return nil
 }
 
 func convertWorkspaces(

coderd/workspaces_test.go

@@ -2872,10 +2872,18 @@ func TestWorkspaceFilterManual(t *testing.T) {
 	t.Run("IncludeAgentMetadata", func(t *testing.T) {
 		t.Parallel()
 
-		client, db := coderdtest.NewWithDatabase(t, nil)
+		// A named non-UTC zone on purpose: jsonb renders timestamptz
+		// in the session TimeZone, and the year-1 collected_at
+		// default renders named zones with LMT second-offsets that Go
+		// refuses to parse unless the query pins UTC.
+		store, ps := dbtestutil.NewDB(t, dbtestutil.WithTimezone("America/Caracas"))
+		client := coderdtest.New(t, &coderdtest.Options{
+			Database: store,
+			Pubsub:   ps,
+		})
 		user := coderdtest.CreateFirstUser(t, client)
 
-		build := dbfake.WorkspaceBuild(t, db, database.WorkspaceTable{
+		build := dbfake.WorkspaceBuild(t, store, database.WorkspaceTable{
 			OrganizationID: user.OrganizationID,
 			OwnerID:        user.UserID,
 		}).WithAgent().Do()
@@ -2887,9 +2895,10 @@ func TestWorkspaceFilterManual(t *testing.T) {
 		collectedAt := dbtime.Now()
 		// Task_Status is mixed-case on purpose: requested keys are
 		// lowercased by the search parser, and the query matches stored
-		// keys case-insensitively.
-		for i, key := range []string{"Task_Status", "cpu", "unrequested"} {
-			err := db.InsertWorkspaceAgentMetadata(ctx, database.InsertWorkspaceAgentMetadataParams{
+		// keys case-insensitively. uncollected is registered but never
+		// reported, so it keeps the year-1 collected_at default.
+		for i, key := range []string{"Task_Status", "cpu", "unrequested", "uncollected"} {
+			err := store.InsertWorkspaceAgentMetadata(ctx, database.InsertWorkspaceAgentMetadataParams{
 				WorkspaceAgentID: agentID,
 				DisplayName:      key,
 				Key:              key,
@@ -2898,10 +2907,13 @@ func TestWorkspaceFilterManual(t *testing.T) {
 				Interval:         int64(time.Second),
 				// Reversed so the response order proves display_order
 				// sorting rather than insertion order.
-				DisplayOrder: int32(3 - i), //nolint:gosec // Tiny test constant.
+				DisplayOrder: int32(4 - i), //nolint:gosec // Tiny test constant.
 			})
 			require.NoError(t, err)
-			err = db.UpdateWorkspaceAgentMetadata(ctx, database.UpdateWorkspaceAgentMetadataParams{
+			if key == "uncollected" {
+				continue
+			}
+			err = store.UpdateWorkspaceAgentMetadata(ctx, database.UpdateWorkspaceAgentMetadataParams{
 				WorkspaceAgentID: agentID,
 				Key:              []string{key},
 				Value:            []string{"value-" + key},
@@ -2927,21 +2939,26 @@ func TestWorkspaceFilterManual(t *testing.T) {
 		require.Empty(t, findAgent(res).Metadata)
 
 		// Opting in returns exactly the requested keys, ordered by
-		// display_order, with their collected values.
+		// display_order, with their collected values. The uncollected
+		// item must round-trip as Go's zero time rather than breaking
+		// the page.
 		res, err = client.Workspaces(reqCtx, codersdk.WorkspaceFilter{
-			IncludeAgentMetadata: []string{"task_status", "cpu"},
+			IncludeAgentMetadata: []string{"task_status", "cpu", "uncollected"},
 		})
 		require.NoError(t, err)
 		metadata := findAgent(res).Metadata
-		require.Len(t, metadata, 2)
-		require.Equal(t, "cpu", metadata[0].Description.Key)
-		require.Equal(t, "value-cpu", metadata[0].Result.Value)
+		require.Len(t, metadata, 3)
+		require.Equal(t, "uncollected", metadata[0].Description.Key)
+		require.Empty(t, metadata[0].Result.Value)
+		require.True(t, metadata[0].Result.CollectedAt.IsZero())
+		require.Equal(t, "cpu", metadata[1].Description.Key)
+		require.Equal(t, "value-cpu", metadata[1].Result.Value)
 		// The collection script is deliberately not exposed on the list
 		// endpoint; it can be long.
-		require.Empty(t, metadata[0].Description.Script)
-		require.Equal(t, "Task_Status", metadata[1].Description.Key)
-		require.Equal(t, "value-Task_Status", metadata[1].Result.Value)
-		require.WithinDuration(t, collectedAt, metadata[1].Result.CollectedAt, time.Second)
+		require.Empty(t, metadata[1].Description.Script)
+		require.Equal(t, "Task_Status", metadata[2].Description.Key)
+		require.Equal(t, "value-Task_Status", metadata[2].Result.Value)
+		require.WithinDuration(t, collectedAt, metadata[2].Result.CollectedAt, time.Second)
 
 		// Unknown keys are not an error; the metadata is just absent.
 		res, err = client.Workspaces(reqCtx, codersdk.WorkspaceFilter{