1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
|
/*-
* Copyright (c) 2014-2015 MongoDB, Inc.
* Copyright (c) 2008-2014 WiredTiger, Inc.
* All rights reserved.
*
* See the file LICENSE for redistribution information.
*/
#include "wt_internal.h"
static int __stat_page(WT_SESSION_IMPL *, WT_PAGE *, WT_DSRC_STATS *);
static void __stat_page_col_var(WT_PAGE *, WT_DSRC_STATS *);
static void __stat_page_row_int(WT_SESSION_IMPL *, WT_PAGE *, WT_DSRC_STATS *);
static void __stat_page_row_leaf(WT_SESSION_IMPL *, WT_PAGE *, WT_DSRC_STATS *);
/*
* __wt_btree_stat_init --
* Initialize the Btree statistics.
*/
int
__wt_btree_stat_init(WT_SESSION_IMPL *session, WT_CURSOR_STAT *cst)
{
WT_BM *bm;
WT_BTREE *btree;
WT_DECL_RET;
WT_DSRC_STATS *stats;
WT_REF *next_walk;
btree = S2BT(session);
bm = btree->bm;
stats = &btree->dhandle->stats;
WT_RET(bm->stat(bm, session, stats));
WT_STAT_SET(stats, btree_fixed_len, btree->bitcnt);
WT_STAT_SET(stats, btree_maximum_depth, btree->maximum_depth);
WT_STAT_SET(stats, btree_maxintlpage, btree->maxintlpage);
WT_STAT_SET(stats, btree_maxintlkey, btree->maxintlkey);
WT_STAT_SET(stats, btree_maxleafpage, btree->maxleafpage);
WT_STAT_SET(stats, btree_maxleafkey, btree->maxleafkey);
WT_STAT_SET(stats, btree_maxleafvalue, btree->maxleafvalue);
/* Everything else is really, really expensive. */
if (!F_ISSET(cst, WT_CONN_STAT_ALL))
return (0);
/*
* Clear the statistics we're about to count.
*/
WT_STAT_SET(stats, btree_column_deleted, 0);
WT_STAT_SET(stats, btree_column_fix, 0);
WT_STAT_SET(stats, btree_column_internal, 0);
WT_STAT_SET(stats, btree_column_variable, 0);
WT_STAT_SET(stats, btree_entries, 0);
WT_STAT_SET(stats, btree_overflow, 0);
WT_STAT_SET(stats, btree_row_internal, 0);
WT_STAT_SET(stats, btree_row_leaf, 0);
next_walk = NULL;
while ((ret = __wt_tree_walk(session, &next_walk, NULL, 0)) == 0 &&
next_walk != NULL) {
WT_WITH_PAGE_INDEX(session,
ret = __stat_page(session, next_walk->page, stats));
WT_RET(ret);
}
return (ret == WT_NOTFOUND ? 0 : ret);
}
/*
* __stat_page --
* Stat any Btree page.
*/
static int
__stat_page(WT_SESSION_IMPL *session, WT_PAGE *page, WT_DSRC_STATS *stats)
{
WT_PAGE_INDEX *pindex;
/*
* All internal pages and overflow pages are trivial, all we track is
* a count of the page type.
*/
switch (page->type) {
case WT_PAGE_COL_FIX:
WT_STAT_INCR(stats, btree_column_fix);
WT_STAT_INCRV(stats, btree_entries, page->pg_fix_entries);
break;
case WT_PAGE_COL_INT:
WT_STAT_INCR(stats, btree_column_internal);
pindex = WT_INTL_INDEX_COPY(page);
WT_STAT_INCRV(stats, btree_entries, pindex->entries);
break;
case WT_PAGE_COL_VAR:
__stat_page_col_var(page, stats);
break;
case WT_PAGE_ROW_INT:
__stat_page_row_int(session, page, stats);
break;
case WT_PAGE_ROW_LEAF:
__stat_page_row_leaf(session, page, stats);
break;
WT_ILLEGAL_VALUE(session);
}
return (0);
}
/*
* __stat_page_col_var --
* Stat a WT_PAGE_COL_VAR page.
*/
static void
__stat_page_col_var(WT_PAGE *page, WT_DSRC_STATS *stats)
{
WT_CELL *cell;
WT_CELL_UNPACK *unpack, _unpack;
WT_COL *cip;
WT_INSERT *ins;
WT_UPDATE *upd;
uint64_t deleted_cnt, entry_cnt, ovfl_cnt;
uint32_t i;
int orig_deleted;
unpack = &_unpack;
deleted_cnt = entry_cnt = ovfl_cnt = 0;
WT_STAT_INCR(stats, btree_column_variable);
/*
* Walk the page counting regular items, adjusting if the item has been
* subsequently deleted or not. This is a mess because 10-item RLE might
* have 3 of the items subsequently deleted. Overflow items are harder,
* we can't know if an updated item will be an overflow item or not; do
* our best, and simply count every overflow item (or RLE set of items)
* we see.
*/
WT_COL_FOREACH(page, cip, i) {
if ((cell = WT_COL_PTR(page, cip)) == NULL) {
orig_deleted = 1;
++deleted_cnt;
} else {
orig_deleted = 0;
__wt_cell_unpack(cell, unpack);
entry_cnt += __wt_cell_rle(unpack);
if (unpack->ovfl)
++ovfl_cnt;
}
/*
* Walk the insert list, checking for changes. For each insert
* we find, correct the original count based on its state.
*/
WT_SKIP_FOREACH(ins, WT_COL_UPDATE(page, cip)) {
upd = ins->upd;
if (WT_UPDATE_DELETED_ISSET(upd)) {
if (!orig_deleted) {
++deleted_cnt;
--entry_cnt;
}
} else
if (orig_deleted) {
--deleted_cnt;
++entry_cnt;
}
}
}
/* Walk any append list. */
WT_SKIP_FOREACH(ins, WT_COL_APPEND(page))
++entry_cnt;
WT_STAT_INCRV(stats, btree_column_deleted, deleted_cnt);
WT_STAT_INCRV(stats, btree_entries, entry_cnt);
WT_STAT_INCRV(stats, btree_overflow, ovfl_cnt);
}
/*
* __stat_page_row_int --
* Stat a WT_PAGE_ROW_INT page.
*/
static void
__stat_page_row_int(
WT_SESSION_IMPL *session, WT_PAGE *page, WT_DSRC_STATS *stats)
{
WT_BTREE *btree;
WT_CELL *cell;
WT_CELL_UNPACK unpack;
WT_PAGE_INDEX *pindex;
uint32_t i, ovfl_cnt;
btree = S2BT(session);
ovfl_cnt = 0;
WT_STAT_INCR(stats, btree_row_internal);
/*
* The number of entries tells us the number of items on row-store
* internal page.
*/
pindex = WT_INTL_INDEX_COPY(page);
WT_STAT_INCRV(stats, btree_entries, pindex->entries);
/*
* Overflow keys are hard: we have to walk the disk image to count them,
* the in-memory representation of the page doesn't necessarily contain
* a reference to the original cell.
*/
if (page->dsk != NULL)
WT_CELL_FOREACH(btree, page->dsk, cell, &unpack, i) {
__wt_cell_unpack(cell, &unpack);
if (__wt_cell_type(cell) == WT_CELL_KEY_OVFL)
++ovfl_cnt;
}
WT_STAT_INCRV(stats, btree_overflow, ovfl_cnt);
}
/*
* __stat_page_row_leaf --
* Stat a WT_PAGE_ROW_LEAF page.
*/
static void
__stat_page_row_leaf(
WT_SESSION_IMPL *session, WT_PAGE *page, WT_DSRC_STATS *stats)
{
WT_BTREE *btree;
WT_CELL *cell;
WT_CELL_UNPACK unpack;
WT_INSERT *ins;
WT_ROW *rip;
WT_UPDATE *upd;
uint32_t entry_cnt, i, ovfl_cnt;
btree = S2BT(session);
entry_cnt = ovfl_cnt = 0;
WT_STAT_INCR(stats, btree_row_leaf);
/*
* Walk any K/V pairs inserted into the page before the first from-disk
* key on the page.
*/
WT_SKIP_FOREACH(ins, WT_ROW_INSERT_SMALLEST(page))
if (!WT_UPDATE_DELETED_ISSET(ins->upd))
++entry_cnt;
/*
* Walk the page's K/V pairs. Count overflow values, where an overflow
* item is any on-disk overflow item that hasn't been updated.
*/
WT_ROW_FOREACH(page, rip, i) {
upd = WT_ROW_UPDATE(page, rip);
if (upd == NULL || !WT_UPDATE_DELETED_ISSET(upd))
++entry_cnt;
if (upd == NULL && (cell =
__wt_row_leaf_value_cell(page, rip, NULL)) != NULL &&
__wt_cell_type(cell) == WT_CELL_VALUE_OVFL)
++ovfl_cnt;
/* Walk K/V pairs inserted after the on-page K/V pair. */
WT_SKIP_FOREACH(ins, WT_ROW_INSERT(page, rip))
if (!WT_UPDATE_DELETED_ISSET(ins->upd))
++entry_cnt;
}
/*
* Overflow keys are hard: we have to walk the disk image to count them,
* the in-memory representation of the page doesn't necessarily contain
* a reference to the original cell.
*/
if (page->dsk != NULL)
WT_CELL_FOREACH(btree, page->dsk, cell, &unpack, i) {
__wt_cell_unpack(cell, &unpack);
if (__wt_cell_type(cell) == WT_CELL_KEY_OVFL)
++ovfl_cnt;
}
WT_STAT_INCRV(stats, btree_entries, entry_cnt);
WT_STAT_INCRV(stats, btree_overflow, ovfl_cnt);
}
|