[receiver/sqlserver] Fix query sample dropping sessions blocked on schema locks

.chloggen/sqlserver-schema-lock-blocking-fix.yaml

@@ -0,0 +1,26 @@
+# Use this changelog template to create an entry for release notes.
+
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: bug_fix
+
+# The name of the component, or a single word describing the area of concern, (e.g. receiver/filelog)
+component: receiver/sqlserver
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: Fix query sample events silently omitting sessions blocked on schema locks.
+
+# Mandatory: One or more tracking issues related to the change. You can use the PR number here if no issue exists.
+issues: [49983]
+
+# (Optional) One or more lines of additional information to render under the primary note.
+# These lines will be padded with 2 spaces and then inserted directly into the document.
+# Use pipe (|) for multiline entries.
+
+# If your change doesn't affect end users or the exported elements of any package,
+# you should instead start your pull request title with [chore] or use the "Skip Changelog" label.
+# Optional: The change log or logs in which this entry should be included.
+# e.g. '[user]' or '[user, api]'
+# Include 'user' if the change is relevant to end users.
+# Include 'api' if there is a change to a library API.
+# Default: '[user]'
+change_logs: [user]

receiver/sqlserverreceiver/queries_test.go

@@ -201,6 +201,66 @@ func TestGetSQLServerQuerySamplesQuery(t *testing.T) {
 	}
 }
 
+// TestQuerySampleQueryDetectsSchemaLockBlocking is a regression test for the class of
+// bugs where sessions blocked on LCK_M_SCH_S / LCK_M_SCH_M (schema stability / modification
+// locks) were silently dropped from the query sample results. The root cause was
+// `CROSS APPLY sys.dm_exec_sql_text(r.plan_handle)`: for a session waiting on a schema
+// lock, `sys.dm_exec_sql_text` returns zero rows, and `CROSS APPLY` therefore eliminates
+// the outer row entirely, hiding the blocking session from downstream reporting.
+//
+// The fix uses `OUTER APPLY` (keeps the row when the TVF returns nothing) and adds
+// `OUTER APPLY sys.dm_exec_input_buffer(...)` as a fallback source for statement_text.
+//
+// This test asserts the invariants that prevent the bug from being re-introduced.
+// If a future change violates any of these, the failing assertion pinpoints the
+// specific problem instead of just a diff error.
+func TestQuerySampleQueryDetectsSchemaLockBlocking(t *testing.T) {
+	query := getSQLServerQuerySamplesQuery()
+
+	// Invariant 1: must not use CROSS APPLY on dm_exec_sql_text.
+	// CROSS APPLY drops rows when the TVF returns zero rows, which is exactly what
+	// happens for sessions blocked on LCK_M_SCH_S / LCK_M_SCH_M.
+	require.NotContains(t, query, "CROSS APPLY sys.dm_exec_sql_text",
+		"query sample must not use CROSS APPLY on sys.dm_exec_sql_text — it silently drops "+
+			"sessions blocked on LCK_M_SCH_S / LCK_M_SCH_M. Use OUTER APPLY instead.")
+
+	// Invariant 2: must use OUTER APPLY on dm_exec_sql_text.
+	require.Contains(t, query, "OUTER APPLY sys.dm_exec_sql_text(r.plan_handle)",
+		"query sample must use OUTER APPLY on sys.dm_exec_sql_text(r.plan_handle) so that "+
+			"sessions with unresolvable plan_handle (e.g. blocked on schema locks) are still emitted.")
+
+	// Invariant 3: must include sys.dm_exec_input_buffer as a fallback source of
+	// statement_text. When plan_handle is unresolvable, dm_exec_input_buffer reads
+	// directly from the client connection buffer and still returns the submitted SQL.
+	require.Contains(t, query, "sys.dm_exec_input_buffer",
+		"query sample must join sys.dm_exec_input_buffer to provide fallback statement_text "+
+			"when plan_handle cannot be resolved.")
+
+	// Invariant 4: WHERE clause must be NULL-safe for o.TEXT.
+	// A bare `WHERE o.TEXT NOT LIKE ...` treats NULL as UNKNOWN and drops the row —
+	// silently re-introducing the same class of bug for encrypted procs and other
+	// cases where dm_exec_sql_text returns a row with NULL text.
+	require.Contains(t, query, "o.TEXT IS NULL OR o.TEXT NOT LIKE",
+		"query sample WHERE clause must be NULL-safe for o.TEXT so that rows with NULL "+
+			"plan text (e.g. encrypted stored procedures) are not silently dropped.")
+
+	// Invariant 5: WHERE clause must be NULL-safe for input_buffer text as well.
+	require.Contains(t, query, "ib.event_info IS NULL OR ib.event_info NOT LIKE",
+		"query sample WHERE clause must be NULL-safe for ib.event_info.")
+
+	// Invariant 6: statement_text must prefer plan_handle-based extraction (which uses
+	// statement_start_offset / statement_end_offset to isolate the *specific* statement
+	// inside a multi-statement batch), and only fall back to input_buffer (which
+	// returns the whole batch) when the plan_handle lookup failed. Using COALESCE
+	// with SUBSTRING first, then ib.event_info, preserves statement-level granularity
+	// for healthy sessions.
+	require.Contains(t, query, "COALESCE(",
+		"query sample must COALESCE plan_handle SUBSTRING with input_buffer fallback so "+
+			"statement-level granularity is preserved for healthy sessions.")
+	require.Contains(t, query, "ib.event_info,",
+		"query sample must reference ib.event_info in the COALESCE fallback chain.")
+}
+
 func TestGetSQLServerIdleBlockingSessionsQuery(t *testing.T) {
 	expectedBytes, err := os.ReadFile(path.Join(".", "testdata", "testIdleBlockerQuerySampleQuery.txt"))
 	require.NoError(t, err)

receiver/sqlserverreceiver/templates/sqlServerQuerySample.tmpl

@@ -10,15 +10,18 @@
   r.STATUS AS request_status,
   ISNULL(s.host_name, '') AS host_name,
   r.command,
-  SUBSTRING(o.TEXT, (r.statement_start_offset / 2) + 1, (
-      (
-        CASE r.statement_end_offset
-          WHEN - 1
-            THEN DATALENGTH(o.TEXT)
-          ELSE r.statement_end_offset
-          END - r.statement_start_offset
-        ) / 2
-      ) + 1) AS statement_text,
+  COALESCE(
+    SUBSTRING(o.TEXT, (r.statement_start_offset / 2) + 1, (
+        (
+          CASE r.statement_end_offset
+            WHEN - 1
+              THEN DATALENGTH(o.TEXT)
+            ELSE r.statement_end_offset
+            END - r.statement_start_offset
+          ) / 2
+        ) + 1),
+    ib.event_info,
+    '') AS statement_text,
   r.blocking_session_id,
   CASE
     WHEN r.blocking_session_id > 0 AND r.wait_time > 0
@@ -56,6 +59,8 @@
 FROM sys.dm_exec_requests r
 INNER JOIN sys.dm_exec_sessions s ON r.session_id = s.session_id
 INNER JOIN sys.dm_exec_connections c ON s.session_id = c.session_id
-CROSS APPLY sys.dm_exec_sql_text(r.plan_handle) AS o
-WHERE o.TEXT NOT LIKE N'/* otel-collector */%'
+OUTER APPLY sys.dm_exec_sql_text(r.plan_handle) AS o
+OUTER APPLY sys.dm_exec_input_buffer(r.session_id, r.request_id) AS ib
+WHERE (o.TEXT IS NULL OR o.TEXT NOT LIKE N'/* otel-collector */%')
+	AND (ib.event_info IS NULL OR ib.event_info NOT LIKE N'/* otel-collector */%')
 	AND r.session_id != @@SPID;

receiver/sqlserverreceiver/testdata/testQuerySampleQuery.txt

@@ -10,15 +10,18 @@
   r.STATUS AS request_status,
   ISNULL(s.host_name, '') AS host_name,
   r.command,
-  SUBSTRING(o.TEXT, (r.statement_start_offset / 2) + 1, (
-      (
-        CASE r.statement_end_offset
-          WHEN - 1
-            THEN DATALENGTH(o.TEXT)
-          ELSE r.statement_end_offset
-          END - r.statement_start_offset
-        ) / 2
-      ) + 1) AS statement_text,
+  COALESCE(
+    SUBSTRING(o.TEXT, (r.statement_start_offset / 2) + 1, (
+        (
+          CASE r.statement_end_offset
+            WHEN - 1
+              THEN DATALENGTH(o.TEXT)
+            ELSE r.statement_end_offset
+            END - r.statement_start_offset
+          ) / 2
+        ) + 1),
+    ib.event_info,
+    '') AS statement_text,
   r.blocking_session_id,
   CASE
     WHEN r.blocking_session_id > 0 AND r.wait_time > 0
@@ -56,6 +59,8 @@
 FROM sys.dm_exec_requests r
 INNER JOIN sys.dm_exec_sessions s ON r.session_id = s.session_id
 INNER JOIN sys.dm_exec_connections c ON s.session_id = c.session_id
-CROSS APPLY sys.dm_exec_sql_text(r.plan_handle) AS o
-WHERE o.TEXT NOT LIKE N'/* otel-collector */%'
+OUTER APPLY sys.dm_exec_sql_text(r.plan_handle) AS o
+OUTER APPLY sys.dm_exec_input_buffer(r.session_id, r.request_id) AS ib
+WHERE (o.TEXT IS NULL OR o.TEXT NOT LIKE N'/* otel-collector */%')
+	AND (ib.event_info IS NULL OR ib.event_info NOT LIKE N'/* otel-collector */%')
 	AND r.session_id != @@SPID;