| 1 |
<!-- doc/src/sgml/pgbuffercache.sgml --> |
| 2 |
|
| 3 |
<sect1 id="pgbuffercache" xreflabel="pg_buffercache"> |
| 4 |
<title>pg_buffercache — inspect <productname>PostgreSQL</productname> |
| 5 |
buffer cache state</title> |
| 6 |
|
| 7 |
<indexterm zone="pgbuffercache"> |
| 8 |
<primary>pg_buffercache</primary> |
| 9 |
</indexterm> |
| 10 |
|
| 11 |
<para> |
| 12 |
The <filename>pg_buffercache</filename> module provides a means for |
| 13 |
examining what's happening in the shared buffer cache in real time. |
| 14 |
It also offers a low-level way to evict data from it, for testing |
| 15 |
purposes. |
| 16 |
</para> |
| 17 |
|
| 18 |
<indexterm> |
| 19 |
<primary>pg_buffercache_pages</primary> |
| 20 |
</indexterm> |
| 21 |
|
| 22 |
<indexterm> |
| 23 |
<primary>pg_buffercache_numa</primary> |
| 24 |
</indexterm> |
| 25 |
|
| 26 |
<indexterm> |
| 27 |
<primary>pg_buffercache_summary</primary> |
| 28 |
</indexterm> |
| 29 |
|
| 30 |
<indexterm> |
| 31 |
<primary>pg_buffercache_usage_counts</primary> |
| 32 |
</indexterm> |
| 33 |
|
| 34 |
<indexterm> |
| 35 |
<primary>pg_buffercache_evict</primary> |
| 36 |
</indexterm> |
| 37 |
|
| 38 |
<indexterm> |
| 39 |
<primary>pg_buffercache_evict_relation</primary> |
| 40 |
</indexterm> |
| 41 |
|
| 42 |
<indexterm> |
| 43 |
<primary>pg_buffercache_evict_all</primary> |
| 44 |
</indexterm> |
| 45 |
|
| 46 |
<para> |
| 47 |
This module provides the <function>pg_buffercache_pages()</function> |
| 48 |
function (wrapped in the <structname>pg_buffercache</structname> view), the |
| 49 |
<function>pg_buffercache_numa_pages()</function> function (wrapped in the |
| 50 |
<structname>pg_buffercache_numa</structname> view), the |
| 51 |
<function>pg_buffercache_summary()</function> function, the |
| 52 |
<function>pg_buffercache_usage_counts()</function> function, the |
| 53 |
<function>pg_buffercache_evict()</function> function, the |
| 54 |
<function>pg_buffercache_evict_relation()</function> function and the |
| 55 |
<function>pg_buffercache_evict_all()</function> function. |
| 56 |
</para> |
| 57 |
|
| 58 |
<para> |
| 59 |
The <function>pg_buffercache_pages()</function> function returns a set of |
| 60 |
records, each row describing the state of one shared buffer entry. The |
| 61 |
<structname>pg_buffercache</structname> view wraps the function for |
| 62 |
convenient use. |
| 63 |
</para> |
| 64 |
|
| 65 |
<para> |
| 66 |
The <function>pg_buffercache_numa_pages()</function> function provides |
| 67 |
<acronym>NUMA</acronym> node mappings for shared buffer entries. This |
| 68 |
information is not part of <function>pg_buffercache_pages()</function> |
| 69 |
itself, as it is much slower to retrieve. |
| 70 |
The <structname>pg_buffercache_numa</structname> view wraps the function for |
| 71 |
convenient use. |
| 72 |
</para> |
| 73 |
|
| 74 |
<para> |
| 75 |
The <function>pg_buffercache_summary()</function> function returns a single |
| 76 |
row summarizing the state of the shared buffer cache. |
| 77 |
</para> |
| 78 |
|
| 79 |
<para> |
| 80 |
The <function>pg_buffercache_usage_counts()</function> function returns a set |
| 81 |
of records, each row describing the number of buffers with a given usage |
| 82 |
count. |
| 83 |
</para> |
| 84 |
|
| 85 |
<para> |
| 86 |
By default, use of the above functions is restricted to superusers and roles |
| 87 |
with privileges of the <literal>pg_monitor</literal> role. Access may be |
| 88 |
granted to others using <command>GRANT</command>. |
| 89 |
</para> |
| 90 |
|
| 91 |
<para> |
| 92 |
The <function>pg_buffercache_evict()</function> function allows a block to |
| 93 |
be evicted from the buffer pool given a buffer identifier. Use of this |
| 94 |
function is restricted to superusers only. |
| 95 |
</para> |
| 96 |
|
| 97 |
<para> |
| 98 |
The <function>pg_buffercache_evict_relation()</function> function allows all |
| 99 |
unpinned shared buffers in the relation to be evicted from the buffer pool |
| 100 |
given a relation identifier. Use of this function is restricted to |
| 101 |
superusers only. |
| 102 |
</para> |
| 103 |
|
| 104 |
<para> |
| 105 |
The <function>pg_buffercache_evict_all()</function> function allows all |
| 106 |
unpinned shared buffers to be evicted in the buffer pool. Use of this |
| 107 |
function is restricted to superusers only. |
| 108 |
</para> |
| 109 |
|
| 110 |
<sect2 id="pgbuffercache-pg-buffercache"> |
| 111 |
<title>The <structname>pg_buffercache</structname> View</title> |
| 112 |
|
| 113 |
<para> |
| 114 |
The definitions of the columns exposed by the view are shown in <xref linkend="pgbuffercache-columns"/>. |
| 115 |
</para> |
| 116 |
|
| 117 |
<table id="pgbuffercache-columns"> |
| 118 |
<title><structname>pg_buffercache</structname> Columns</title> |
| 119 |
<tgroup cols="1"> |
| 120 |
<thead> |
| 121 |
<row> |
| 122 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 123 |
Column Type |
| 124 |
</para> |
| 125 |
<para> |
| 126 |
Description |
| 127 |
</para></entry> |
| 128 |
</row> |
| 129 |
</thead> |
| 130 |
|
| 131 |
<tbody> |
| 132 |
<row> |
| 133 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 134 |
<structfield>bufferid</structfield> <type>integer</type> |
| 135 |
</para> |
| 136 |
<para> |
| 137 |
ID, in the range 1..<varname>shared_buffers</varname> |
| 138 |
</para></entry> |
| 139 |
</row> |
| 140 |
|
| 141 |
<row> |
| 142 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 143 |
<structfield>relfilenode</structfield> <type>oid</type> |
| 144 |
(references <link linkend="catalog-pg-class"><structname>pg_class</structname></link>.<structfield>relfilenode</structfield>) |
| 145 |
</para> |
| 146 |
<para> |
| 147 |
Filenode number of the relation |
| 148 |
</para></entry> |
| 149 |
</row> |
| 150 |
|
| 151 |
<row> |
| 152 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 153 |
<structfield>reltablespace</structfield> <type>oid</type> |
| 154 |
(references <link linkend="catalog-pg-tablespace"><structname>pg_tablespace</structname></link>.<structfield>oid</structfield>) |
| 155 |
</para> |
| 156 |
<para> |
| 157 |
Tablespace OID of the relation |
| 158 |
</para></entry> |
| 159 |
</row> |
| 160 |
|
| 161 |
<row> |
| 162 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 163 |
<structfield>reldatabase</structfield> <type>oid</type> |
| 164 |
(references <link linkend="catalog-pg-database"><structname>pg_database</structname></link>.<structfield>oid</structfield>) |
| 165 |
</para> |
| 166 |
<para> |
| 167 |
Database OID of the relation |
| 168 |
</para></entry> |
| 169 |
</row> |
| 170 |
|
| 171 |
<row> |
| 172 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 173 |
<structfield>relforknumber</structfield> <type>smallint</type> |
| 174 |
</para> |
| 175 |
<para> |
| 176 |
Fork number within the relation; see |
| 177 |
<filename>common/relpath.h</filename> |
| 178 |
</para></entry> |
| 179 |
</row> |
| 180 |
|
| 181 |
<row> |
| 182 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 183 |
<structfield>relblocknumber</structfield> <type>bigint</type> |
| 184 |
</para> |
| 185 |
<para> |
| 186 |
Page number within the relation |
| 187 |
</para></entry> |
| 188 |
</row> |
| 189 |
|
| 190 |
<row> |
| 191 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 192 |
<structfield>isdirty</structfield> <type>boolean</type> |
| 193 |
</para> |
| 194 |
<para> |
| 195 |
Is the page dirty? |
| 196 |
</para></entry> |
| 197 |
</row> |
| 198 |
|
| 199 |
<row> |
| 200 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 201 |
<structfield>usagecount</structfield> <type>smallint</type> |
| 202 |
</para> |
| 203 |
<para> |
| 204 |
Clock-sweep access count |
| 205 |
</para></entry> |
| 206 |
</row> |
| 207 |
|
| 208 |
<row> |
| 209 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 210 |
<structfield>pinning_backends</structfield> <type>integer</type> |
| 211 |
</para> |
| 212 |
<para> |
| 213 |
Number of backends pinning this buffer |
| 214 |
</para></entry> |
| 215 |
</row> |
| 216 |
</tbody> |
| 217 |
</tgroup> |
| 218 |
</table> |
| 219 |
|
| 220 |
<para> |
| 221 |
There is one row for each buffer in the shared cache. Unused buffers are |
| 222 |
shown with all fields null except <structfield>bufferid</structfield>. Shared system |
| 223 |
catalogs are shown as belonging to database zero. |
| 224 |
</para> |
| 225 |
|
| 226 |
<para> |
| 227 |
Because the cache is shared by all the databases, there will normally be |
| 228 |
pages from relations not belonging to the current database. This means |
| 229 |
that there may not be matching join rows in <structname>pg_class</structname> for |
| 230 |
some rows, or that there could even be incorrect joins. If you are |
| 231 |
trying to join against <structname>pg_class</structname>, it's a good idea to |
| 232 |
restrict the join to rows having <structfield>reldatabase</structfield> equal to |
| 233 |
the current database's OID or zero. |
| 234 |
</para> |
| 235 |
|
| 236 |
<para> |
| 237 |
Since buffer manager locks are not taken to copy the buffer state data that |
| 238 |
the view will display, accessing <structname>pg_buffercache</structname> view |
| 239 |
has less impact on normal buffer activity but it doesn't provide a consistent |
| 240 |
set of results across all buffers. However, we ensure that the information of |
| 241 |
each buffer is self-consistent. |
| 242 |
</para> |
| 243 |
</sect2> |
| 244 |
|
| 245 |
<sect2 id="pgbuffercache-pg-buffercache-numa"> |
| 246 |
<title>The <structname>pg_buffercache_numa</structname> View</title> |
| 247 |
|
| 248 |
<para> |
| 249 |
The definitions of the columns exposed by the view are shown in <xref linkend="pgbuffercache-numa-columns"/>. |
| 250 |
</para> |
| 251 |
|
| 252 |
<table id="pgbuffercache-numa-columns"> |
| 253 |
<title><structname>pg_buffercache_numa</structname> Columns</title> |
| 254 |
<tgroup cols="1"> |
| 255 |
<thead> |
| 256 |
<row> |
| 257 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 258 |
Column Type |
| 259 |
</para> |
| 260 |
<para> |
| 261 |
Description |
| 262 |
</para></entry> |
| 263 |
</row> |
| 264 |
</thead> |
| 265 |
|
| 266 |
<tbody> |
| 267 |
<row> |
| 268 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 269 |
<structfield>bufferid</structfield> <type>integer</type> |
| 270 |
</para> |
| 271 |
<para> |
| 272 |
ID, in the range 1..<varname>shared_buffers</varname> |
| 273 |
</para></entry> |
| 274 |
</row> |
| 275 |
|
| 276 |
<row> |
| 277 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 278 |
<structfield>os_page_num</structfield> <type>bigint</type> |
| 279 |
</para> |
| 280 |
<para> |
| 281 |
number of OS memory page for this buffer |
| 282 |
</para></entry> |
| 283 |
</row> |
| 284 |
|
| 285 |
<row> |
| 286 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 287 |
<structfield>numa_node</structfield> <type>int</type> |
| 288 |
</para> |
| 289 |
<para> |
| 290 |
ID of <acronym>NUMA</acronym> node |
| 291 |
</para></entry> |
| 292 |
</row> |
| 293 |
|
| 294 |
</tbody> |
| 295 |
</tgroup> |
| 296 |
</table> |
| 297 |
|
| 298 |
<para> |
| 299 |
As <acronym>NUMA</acronym> node ID inquiry for each page requires memory pages |
| 300 |
to be paged-in, the first execution of this function can take a noticeable |
| 301 |
amount of time. In all the cases (first execution or not), retrieving this |
| 302 |
information is costly and querying the view at a high frequency is not recommended. |
| 303 |
</para> |
| 304 |
|
| 305 |
<warning> |
| 306 |
<para> |
| 307 |
When determining the <acronym>NUMA</acronym> node, the view touches |
| 308 |
all memory pages for the shared memory segment. This will force |
| 309 |
allocation of the shared memory, if it wasn't allocated already, |
| 310 |
and the memory may get allocated in a single <acronym>NUMA</acronym> |
| 311 |
node (depending on system configuration). |
| 312 |
</para> |
| 313 |
</warning> |
| 314 |
|
| 315 |
</sect2> |
| 316 |
|
| 317 |
<sect2 id="pgbuffercache-summary"> |
| 318 |
<title>The <function>pg_buffercache_summary()</function> Function</title> |
| 319 |
|
| 320 |
<para> |
| 321 |
The definitions of the columns exposed by the function are shown in <xref linkend="pgbuffercache-summary-columns"/>. |
| 322 |
</para> |
| 323 |
|
| 324 |
<table id="pgbuffercache-summary-columns"> |
| 325 |
<title><function>pg_buffercache_summary()</function> Output Columns</title> |
| 326 |
<tgroup cols="1"> |
| 327 |
<thead> |
| 328 |
<row> |
| 329 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 330 |
Column Type |
| 331 |
</para> |
| 332 |
<para> |
| 333 |
Description |
| 334 |
</para></entry> |
| 335 |
</row> |
| 336 |
</thead> |
| 337 |
|
| 338 |
<tbody> |
| 339 |
<row> |
| 340 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 341 |
<structfield>buffers_used</structfield> <type>int4</type> |
| 342 |
</para> |
| 343 |
<para> |
| 344 |
Number of used shared buffers |
| 345 |
</para></entry> |
| 346 |
</row> |
| 347 |
|
| 348 |
<row> |
| 349 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 350 |
<structfield>buffers_unused</structfield> <type>int4</type> |
| 351 |
</para> |
| 352 |
<para> |
| 353 |
Number of unused shared buffers |
| 354 |
</para></entry> |
| 355 |
</row> |
| 356 |
|
| 357 |
<row> |
| 358 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 359 |
<structfield>buffers_dirty</structfield> <type>int4</type> |
| 360 |
</para> |
| 361 |
<para> |
| 362 |
Number of dirty shared buffers |
| 363 |
</para></entry> |
| 364 |
</row> |
| 365 |
|
| 366 |
<row> |
| 367 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 368 |
<structfield>buffers_pinned</structfield> <type>int4</type> |
| 369 |
</para> |
| 370 |
<para> |
| 371 |
Number of pinned shared buffers |
| 372 |
</para></entry> |
| 373 |
</row> |
| 374 |
|
| 375 |
<row> |
| 376 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 377 |
<structfield>usagecount_avg</structfield> <type>float8</type> |
| 378 |
</para> |
| 379 |
<para> |
| 380 |
Average usage count of used shared buffers |
| 381 |
</para></entry> |
| 382 |
</row> |
| 383 |
</tbody> |
| 384 |
</tgroup> |
| 385 |
</table> |
| 386 |
|
| 387 |
<para> |
| 388 |
The <function>pg_buffercache_summary()</function> function returns a |
| 389 |
single row summarizing the state of all shared buffers. Similar and more |
| 390 |
detailed information is provided by the |
| 391 |
<structname>pg_buffercache</structname> view, but |
| 392 |
<function>pg_buffercache_summary()</function> is significantly cheaper. |
| 393 |
</para> |
| 394 |
|
| 395 |
<para> |
| 396 |
Like the <structname>pg_buffercache</structname> view, |
| 397 |
<function>pg_buffercache_summary()</function> does not acquire buffer |
| 398 |
manager locks. Therefore concurrent activity can lead to minor inaccuracies |
| 399 |
in the result. |
| 400 |
</para> |
| 401 |
</sect2> |
| 402 |
|
| 403 |
<sect2 id="pgbuffercache-usage-counts"> |
| 404 |
<title>The <function>pg_buffercache_usage_counts()</function> Function</title> |
| 405 |
|
| 406 |
<para> |
| 407 |
The definitions of the columns exposed by the function are shown in |
| 408 |
<xref linkend="pgbuffercache_usage_counts-columns"/>. |
| 409 |
</para> |
| 410 |
|
| 411 |
<table id="pgbuffercache_usage_counts-columns"> |
| 412 |
<title><function>pg_buffercache_usage_counts()</function> Output Columns</title> |
| 413 |
<tgroup cols="1"> |
| 414 |
<thead> |
| 415 |
<row> |
| 416 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 417 |
Column Type |
| 418 |
</para> |
| 419 |
<para> |
| 420 |
Description |
| 421 |
</para></entry> |
| 422 |
</row> |
| 423 |
</thead> |
| 424 |
|
| 425 |
<tbody> |
| 426 |
<row> |
| 427 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 428 |
<structfield>usage_count</structfield> <type>int4</type> |
| 429 |
</para> |
| 430 |
<para> |
| 431 |
A possible buffer usage count |
| 432 |
</para></entry> |
| 433 |
</row> |
| 434 |
|
| 435 |
<row> |
| 436 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 437 |
<structfield>buffers</structfield> <type>int4</type> |
| 438 |
</para> |
| 439 |
<para> |
| 440 |
Number of buffers with the usage count |
| 441 |
</para></entry> |
| 442 |
</row> |
| 443 |
|
| 444 |
<row> |
| 445 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 446 |
<structfield>dirty</structfield> <type>int4</type> |
| 447 |
</para> |
| 448 |
<para> |
| 449 |
Number of dirty buffers with the usage count |
| 450 |
</para></entry> |
| 451 |
</row> |
| 452 |
|
| 453 |
<row> |
| 454 |
<entry role="catalog_table_entry"><para role="column_definition"> |
| 455 |
<structfield>pinned</structfield> <type>int4</type> |
| 456 |
</para> |
| 457 |
<para> |
| 458 |
Number of pinned buffers with the usage count |
| 459 |
</para></entry> |
| 460 |
</row> |
| 461 |
</tbody> |
| 462 |
</tgroup> |
| 463 |
</table> |
| 464 |
|
| 465 |
<para> |
| 466 |
The <function>pg_buffercache_usage_counts()</function> function returns a |
| 467 |
set of rows summarizing the states of all shared buffers, aggregated over |
| 468 |
the possible usage count values. Similar and more detailed information is |
| 469 |
provided by the <structname>pg_buffercache</structname> view, but |
| 470 |
<function>pg_buffercache_usage_counts()</function> is significantly cheaper. |
| 471 |
</para> |
| 472 |
|
| 473 |
<para> |
| 474 |
Like the <structname>pg_buffercache</structname> view, |
| 475 |
<function>pg_buffercache_usage_counts()</function> does not acquire buffer |
| 476 |
manager locks. Therefore concurrent activity can lead to minor inaccuracies |
| 477 |
in the result. |
| 478 |
</para> |
| 479 |
</sect2> |
| 480 |
|
| 481 |
<sect2 id="pgbuffercache-pg-buffercache-evict"> |
| 482 |
<title>The <function>pg_buffercache_evict()</function> Function</title> |
| 483 |
<para> |
| 484 |
The <function>pg_buffercache_evict()</function> function takes a buffer |
| 485 |
identifier, as shown in the <structfield>bufferid</structfield> column of |
| 486 |
the <structname>pg_buffercache</structname> view. It returns information |
| 487 |
about whether the buffer was evicted and flushed. The buffer_evicted |
| 488 |
column is true on success, and false if the buffer wasn't valid, if it |
| 489 |
couldn't be evicted because it was pinned, or if it became dirty again |
| 490 |
after an attempt to write it out. The buffer_flushed column is true if the |
| 491 |
buffer was flushed. This does not necessarily mean that buffer was flushed |
| 492 |
by us, it might be flushed by someone else. The result is immediately out |
| 493 |
of date upon return, as the buffer might become valid again at any time due |
| 494 |
to concurrent activity. The function is intended for developer testing |
| 495 |
only. |
| 496 |
</para> |
| 497 |
</sect2> |
| 498 |
|
| 499 |
<sect2 id="pgbuffercache-pg-buffercache-evict-relation"> |
| 500 |
<title>The <function>pg_buffercache_evict_relation()</function> Function</title> |
| 501 |
<para> |
| 502 |
The <function>pg_buffercache_evict_relation()</function> function is very |
| 503 |
similar to the <function>pg_buffercache_evict()</function> function. The |
| 504 |
difference is that the <function>pg_buffercache_evict_relation()</function> |
| 505 |
takes a relation identifier instead of buffer identifier. It tries to |
| 506 |
evict all buffers for all forks in that relation. |
| 507 |
|
| 508 |
It returns the number of evicted buffers, flushed buffers and the number of |
| 509 |
buffers that could not be evicted. Flushed buffers haven't necessarily |
| 510 |
been flushed by us, they might have been flushed by someone else. The |
| 511 |
result is immediately out of date upon return, as buffers might immediately |
| 512 |
be read back in due to concurrent activity. The function is intended for |
| 513 |
developer testing only. |
| 514 |
</para> |
| 515 |
</sect2> |
| 516 |
|
| 517 |
<sect2 id="pgbuffercache-pg-buffercache-evict-all"> |
| 518 |
<title>The <function>pg_buffercache_evict_all()</function> Function</title> |
| 519 |
<para> |
| 520 |
The <function>pg_buffercache_evict_all()</function> function is very |
| 521 |
similar to the <function>pg_buffercache_evict()</function> function. The |
| 522 |
difference is, the <function>pg_buffercache_evict_all()</function> function |
| 523 |
does not take an argument; instead it tries to evict all buffers in the |
| 524 |
buffer pool. It returns the number of evicted buffers, flushed buffers and |
| 525 |
the number of buffers that could not be evicted. Flushed buffers haven't |
| 526 |
necessarily been flushed by us, they might have been flushed by someone |
| 527 |
else. The result is immediately out of date upon return, as buffers might |
| 528 |
immediately be read back in due to concurrent activity. The function is |
| 529 |
intended for developer testing only. |
| 530 |
</para> |
| 531 |
</sect2> |
| 532 |
|
| 533 |
<sect2 id="pgbuffercache-sample-output"> |
| 534 |
<title>Sample Output</title> |
| 535 |
|
| 536 |
<screen> |
| 537 |
regression=# SELECT n.nspname, c.relname, count(*) AS buffers |
| 538 |
FROM pg_buffercache b JOIN pg_class c |
| 539 |
ON b.relfilenode = pg_relation_filenode(c.oid) AND |
| 540 |
b.reldatabase IN (0, (SELECT oid FROM pg_database |
| 541 |
WHERE datname = current_database())) |
| 542 |
JOIN pg_namespace n ON n.oid = c.relnamespace |
| 543 |
GROUP BY n.nspname, c.relname |
| 544 |
ORDER BY 3 DESC |
| 545 |
LIMIT 10; |
| 546 |
|
| 547 |
nspname | relname | buffers |
| 548 |
------------+------------------------+--------- |
| 549 |
public | delete_test_table | 593 |
| 550 |
public | delete_test_table_pkey | 494 |
| 551 |
pg_catalog | pg_attribute | 472 |
| 552 |
public | quad_poly_tbl | 353 |
| 553 |
public | tenk2 | 349 |
| 554 |
public | tenk1 | 349 |
| 555 |
public | gin_test_idx | 306 |
| 556 |
pg_catalog | pg_largeobject | 206 |
| 557 |
public | gin_test_tbl | 188 |
| 558 |
public | spgist_text_tbl | 182 |
| 559 |
(10 rows) |
| 560 |
|
| 561 |
|
| 562 |
regression=# SELECT * FROM pg_buffercache_summary(); |
| 563 |
buffers_used | buffers_unused | buffers_dirty | buffers_pinned | usagecount_avg |
| 564 |
--------------+----------------+---------------+----------------+---------------- |
| 565 |
248 | 2096904 | 39 | 0 | 3.141129 |
| 566 |
(1 row) |
| 567 |
|
| 568 |
|
| 569 |
regression=# SELECT * FROM pg_buffercache_usage_counts(); |
| 570 |
usage_count | buffers | dirty | pinned |
| 571 |
-------------+---------+-------+-------- |
| 572 |
0 | 14650 | 0 | 0 |
| 573 |
1 | 1436 | 671 | 0 |
| 574 |
2 | 102 | 88 | 0 |
| 575 |
3 | 23 | 21 | 0 |
| 576 |
4 | 9 | 7 | 0 |
| 577 |
5 | 164 | 106 | 0 |
| 578 |
(6 rows) |
| 579 |
</screen> |
| 580 |
</sect2> |
| 581 |
|
| 582 |
<sect2 id="pgbuffercache-authors"> |
| 583 |
<title>Authors</title> |
| 584 |
|
| 585 |
<para> |
| 586 |
Mark Kirkwood <email>[email protected]</email> |
| 587 |
</para> |
| 588 |
|
| 589 |
<para> |
| 590 |
Design suggestions: Neil Conway <email>[email protected]</email> |
| 591 |
</para> |
| 592 |
|
| 593 |
<para> |
| 594 |
Debugging advice: Tom Lane <email>[email protected]</email> |
| 595 |
</para> |
| 596 |
</sect2> |
| 597 |
|
| 598 |
</sect1> |