From fd45f737ccbdde1c69b4644e0d3da16f181952cc Mon Sep 17 00:00:00 2001
From: Yifan Xu <30385241+xuyifangreeneyes@users.noreply.github.com>
Date: Fri, 10 Mar 2023 14:15:13 +0800
Subject: [PATCH] statistics: fix wrong column stats loading after analyze
 twice (#42076)

close pingcap/tidb#42073
---
 statistics/integration_test.go      | 26 ++++++++++++++++++++++++++
 statistics/interact_with_storage.go | 15 ++++++++++++++-
 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/statistics/integration_test.go b/statistics/integration_test.go
index 1c8672b790..569a23dbbb 100644
--- a/statistics/integration_test.go
+++ b/statistics/integration_test.go
@@ -693,3 +693,29 @@ func TestSingleColumnIndexNDV(t *testing.T) {
 		require.Equal(t, expectedResults[i][2], row[7]) // null_count
 	}
 }
+
+func TestColumnStatsLazyLoad(t *testing.T) {
+	store, dom := testkit.CreateMockStoreAndDomain(t)
+	tk := testkit.NewTestKit(t, store)
+	h := dom.StatsHandle()
+	originLease := h.Lease()
+	defer h.SetLease(originLease)
+	// Set `Lease` to `Millisecond` to enable column stats lazy load.
+	h.SetLease(time.Millisecond)
+	tk.MustExec("use test")
+	tk.MustExec("create table t(a int, b int)")
+	tk.MustExec("insert into t values (1,2), (3,4), (5,6), (7,8)")
+	require.NoError(t, h.HandleDDLEvent(<-h.DDLEventCh()))
+	tk.MustExec("analyze table t")
+	is := dom.InfoSchema()
+	tbl, err := is.TableByName(model.NewCIStr("test"), model.NewCIStr("t"))
+	require.NoError(t, err)
+	tblInfo := tbl.Meta()
+	c1 := tblInfo.Columns[0]
+	c2 := tblInfo.Columns[1]
+	require.True(t, h.GetTableStats(tblInfo).Columns[c1.ID].IsAllEvicted())
+	require.True(t, h.GetTableStats(tblInfo).Columns[c2.ID].IsAllEvicted())
+	tk.MustExec("analyze table t")
+	require.True(t, h.GetTableStats(tblInfo).Columns[c1.ID].IsAllEvicted())
+	require.True(t, h.GetTableStats(tblInfo).Columns[c2.ID].IsAllEvicted())
+}
diff --git a/statistics/interact_with_storage.go b/statistics/interact_with_storage.go
index c0acce31b7..1ad1b82080 100644
--- a/statistics/interact_with_storage.go
+++ b/statistics/interact_with_storage.go
@@ -334,10 +334,23 @@ func columnStatsFromStorage(reader *StatsReader, row chunk.Row, table *Table, ta
 		// 2. this column is not handle, and:
 		// 3. the column doesn't has any statistics before, and:
 		// 4. loadAll is false.
+		//
+		// Here is the explanation of the condition `!col.IsStatsInitialized() || col.IsAllEvicted()`.
+		// For one column:
+		// 1. If there is no stats for it in the storage(i.e., analyze has never been executed before), then its stats status
+		//    would be `!col.IsStatsInitialized()`. In this case we should go the `notNeedLoad` path.
+		// 2. If there exists stats for it in the storage but its stats status is `col.IsAllEvicted()`, there are two
+		//    sub cases for this case. One is that the column stats have never been used/needed by the optimizer so they have
+		//    never been loaded. The other is that the column stats were loaded and then evicted. For the both sub cases,
+		//    we should go the `notNeedLoad` path.
+		// 3. If some parts(Histogram/TopN/CMSketch) of stats for it exist in TiDB memory currently, we choose to load all of
+		//    its new stats once we find stats version is updated.
 		notNeedLoad := lease > 0 &&
 			!isHandle &&
-			(col == nil || !col.IsStatsInitialized() && col.LastUpdateVersion < histVer) &&
+			(col == nil || ((!col.IsStatsInitialized() || col.IsAllEvicted()) && col.LastUpdateVersion < histVer)) &&
 			!loadAll
+		// Here is
+		//For one column, if there is no stats for it in the storage(analyze is never)
 		if notNeedLoad {
 			count, err := ColumnCountFromStorage(reader, table.PhysicalID, histID, statsVer)
 			if err != nil {