mirror of
https://github.com/MariaDB/server.git
synced 2025-07-18 09:18:15 +02:00

We deprecate and ignore the parameter innodb_buffer_pool_chunk_size and let the buffer pool size to be changed in arbitrary 1-megabyte increments. innodb_buffer_pool_size_max: A new read-only startup parameter that specifies the maximum innodb_buffer_pool_size. If 0 or unspecified, it will default to the specified innodb_buffer_pool_size rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems. This maximum is very likely to be limited further by the operating system. The status variable Innodb_buffer_pool_resize_status will reflect the status of shrinking the buffer pool. When no shrinking is in progress, the string will be empty. Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size will block until the requested buffer pool size change has been implemented, or the execution is interrupted by a KILL statement a client disconnect, or server shutdown. If the buf_flush_page_cleaner() thread notices that we are running out of memory, the operation may fail with ER_WRONG_USAGE. SET GLOBAL innodb_buffer_pool_size will be refused if the server was started with --large-pages (even if no HugeTLB pages were successfully allocated). This functionality is somewhat exercised by the test main.large_pages, which now runs also on Microsoft Windows. On Linux, explicit HugeTLB mappings are apparently excluded from the reported Redident Set Size (RSS), and apparently unshrinkable between mmap(2) and munmap(2). The buffer pool will be mapped to a contiguous virtual memory area that will be aligned and partitioned into extents of 8 MiB on 64-bit systems and 2 MiB on 32-bit systems. Within an extent, the first few innodb_page_size blocks contain buf_block_t objects that will cover the page frames in the rest of the extent. The number of such frames is precomputed in the array first_page_in_extent[] for each innodb_page_size. In this way, there is a trivial mapping between page frames and block descriptors and we do not need any lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map. We will always allocate the same number of block descriptors for an extent, even if we do not need all the buf_block_t in the last extent in case the innodb_buffer_pool_size is not an integer multiple of the of extents size. The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default innodb_page_size=16k this corresponds to 5 MiB. However, now that the innodb_buffer_pool_size includes the memory allocated for the block descriptors, the minimum would be innodb_buffer_pool_size=6m. my_large_virtual_alloc(): A new function, similar to my_large_malloc(). my_virtual_mem_reserve(), my_virtual_mem_commit(), my_virtual_mem_decommit(), my_virtual_mem_release(): New interface mostly by Vladislav Vaintroub, to separately reserve and release virtual address space, as well as to commit and decommit memory within it. After my_virtual_mem_decommit(), the virtual memory range will be read-only or unaccessible, depending on whether the build option cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1 has been specified. This option is hard-coded on Microsoft Windows, where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible. On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory will be zeroed out immediately. On other POSIX-like systems, madvise(MADV_FREE) will be used if available, to give the operating system kernel a permission to zero out the virtual memory range. We prefer immediate freeing so that the reported resident set size (RSS) of the process will reflect the current innodb_buffer_pool_size. Shrinking the buffer pool is a rarely executed resource intensive operation, and the immediate configuration of the MMU mappings should not incur significant additional penalty. opt_super_large_pages: Declare only on Solaris. Actually, this is specific to the SPARC implementation of Solaris, but because we lack access to a Solaris development environment, we will not revise this for other MMU and ISA. buf_pool_t::chunk_t::create(): Remove. buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list. buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only(). buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>, only to be modified by the buf_flush_page_cleaner() thread. buf_pool_t::shrink(): Attempt to shrink the buffer pool. There are 3 possible outcomes: SHRINK_DONE (success), SHRINK_IN_PROGRESS (the caller may keep trying), and SHRINK_ABORT (we seem to be running out of buffer pool). While traversing buf_pool.LRU, release the contended buf_pool.mutex once in every 32 iterations in order to reduce starvation. Use lru_scan_itr for efficient traversal, similar to buf_LRU_free_from_common_LRU_list(). buf_pool_t::shrunk(): Update the reduced size of the buffer pool in a way that is compatible with buf_pool_t::page_guess(), and invoke my_virtual_mem_decommit(). buf_pool_t::resize(): Before invoking shrink(), run one batch of buf_flush_page_cleaner() in order to prevent LRU_warn(). Abort if shrink() recommends it, or no blocks were withdrawn in the past 15 seconds, or the execution of the statement SET GLOBAL innodb_buffer_pool_size was interrupted. buf_pool_t::first_to_withdraw: The first block descriptor that is out of the bounds of the shrunk buffer pool. buf_pool_t::withdrawn: The list of withdrawn blocks. If buf_pool_t::resize() is aborted before shrink() completes, we must be able to resurrect the withdrawn blocks in the free list. buf_pool_t::contains_zip(): Added a parameter for the number of least significant pointer bits to disregard, so that we can find any pointers to within a block that is supposed to be free. buf_pool_t::is_shrinking(): Return the total number or blocks that were withdrawn or are to be withdrawn. buf_pool_t::to_withdraw(): Return the number of blocks that will need to be withdrawn. buf_pool_t::usable_size(): Number of usable pages, considering possible in-progress attempt at shrinking the buffer pool. buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer. If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will be validated before being dereferenced. buf_pool_t::get_info(): Replaces buf_stats_get_pool_info(). innodb_init_param(): Refactored. We must first compute srv_page_size_shift and then determine the valid bounds of innodb_buffer_pool_size. buf_buddy_shrink(): Replaces buf_buddy_realloc(). Part of the work is deferred to buf_buddy_condense_free(), which is being executed when we are not holding any buf_pool.page_hash latch. buf_buddy_condense_free(): Do not relocate blocks. buf_buddy_free_low(): Do not care about buffer pool shrinking. This will be handled by buf_buddy_shrink() and buf_buddy_condense_free(). buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip() when we are allocating from the binary buddy system. Previously we were asserting this on multiple recursion levels. buf_buddy_block_free(), buf_buddy_free_low(): Assert !buf_pool.contains_zip(). buf_buddy_alloc_from(): Remove the redundant parameter j. buf_flush_LRU_list_batch(): Add the parameter to_withdraw to keep track of buf_pool.n_blocks_to_withdraw. buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch() if we are shrinking the buffer pool. In that case, we want to minimize the page relocations and just finish as quickly as possible. trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled() in every iteration, in case the buffer pool is being shrunk in the middle of a purge batch. Reviewed by: Debarun Banerjee
779 lines
20 KiB
C++
779 lines
20 KiB
C++
/*****************************************************************************
|
|
|
|
Copyright (c) 2011, 2017, Oracle and/or its affiliates. All Rights Reserved.
|
|
Copyright (c) 2017, 2022, MariaDB Corporation.
|
|
|
|
This program is free software; you can redistribute it and/or modify it under
|
|
the terms of the GNU General Public License as published by the Free Software
|
|
Foundation; version 2 of the License.
|
|
|
|
This program is distributed in the hope that it will be useful, but WITHOUT
|
|
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
|
|
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
|
|
|
|
You should have received a copy of the GNU General Public License along with
|
|
this program; if not, write to the Free Software Foundation, Inc.,
|
|
51 Franklin Street, Fifth Floor, Boston, MA 02110-1335 USA
|
|
|
|
*****************************************************************************/
|
|
|
|
/**************************************************//**
|
|
@file buf/buf0dump.cc
|
|
Implements a buffer pool dump/load.
|
|
|
|
Created April 08, 2011 Vasil Dimov
|
|
*******************************************************/
|
|
|
|
#include "my_global.h"
|
|
#include "mysqld.h"
|
|
#include "my_sys.h"
|
|
|
|
#include "mysql/psi/mysql_stage.h"
|
|
#include "mysql/psi/psi.h"
|
|
|
|
#include "buf0rea.h"
|
|
#include "buf0dump.h"
|
|
#include "dict0load.h"
|
|
#include "os0file.h"
|
|
#include "srv0srv.h"
|
|
#include "srv0start.h"
|
|
#include "ut0byte.h"
|
|
|
|
#include <algorithm>
|
|
|
|
#include "mysql/service_wsrep.h" /* wsrep_recovery */
|
|
#include <my_service_manager.h>
|
|
|
|
static void buf_do_load_dump();
|
|
|
|
enum status_severity {
|
|
STATUS_INFO,
|
|
STATUS_ERR
|
|
};
|
|
|
|
#define SHUTTING_DOWN() (srv_shutdown_state != SRV_SHUTDOWN_NONE)
|
|
|
|
/* Flags that tell the buffer pool dump/load thread which action should it
|
|
take after being waked up. */
|
|
static volatile bool buf_dump_should_start;
|
|
static volatile bool buf_load_should_start;
|
|
|
|
static Atomic_relaxed<bool> buf_load_abort_flag;
|
|
|
|
/** Start the buffer pool dump/load task and instructs it to start a dump. */
|
|
void buf_dump_start()
|
|
{
|
|
buf_dump_should_start= true;
|
|
buf_do_load_dump();
|
|
}
|
|
|
|
/** Start the buffer pool dump/load task and instructs it to start a load. */
|
|
void buf_load_start()
|
|
{
|
|
buf_load_should_start= true;
|
|
buf_do_load_dump();
|
|
}
|
|
|
|
/*****************************************************************//**
|
|
Sets the global variable that feeds MySQL's innodb_buffer_pool_dump_status
|
|
to the specified string. The format and the following parameters are the
|
|
same as the ones used for printf(3). The value of this variable can be
|
|
retrieved by:
|
|
SELECT variable_value FROM information_schema.global_status WHERE
|
|
variable_name = 'INNODB_BUFFER_POOL_DUMP_STATUS';
|
|
or by:
|
|
SHOW STATUS LIKE 'innodb_buffer_pool_dump_status'; */
|
|
static MY_ATTRIBUTE((nonnull, format(printf, 2, 3)))
|
|
void
|
|
buf_dump_status(
|
|
/*============*/
|
|
enum status_severity severity,/*!< in: status severity */
|
|
const char* fmt, /*!< in: format */
|
|
...) /*!< in: extra parameters according
|
|
to fmt */
|
|
{
|
|
va_list ap;
|
|
|
|
va_start(ap, fmt);
|
|
|
|
vsnprintf(
|
|
export_vars.innodb_buffer_pool_dump_status,
|
|
sizeof(export_vars.innodb_buffer_pool_dump_status),
|
|
fmt, ap);
|
|
|
|
switch (severity) {
|
|
case STATUS_INFO:
|
|
ib::info() << export_vars.innodb_buffer_pool_dump_status;
|
|
break;
|
|
|
|
case STATUS_ERR:
|
|
ib::error() << export_vars.innodb_buffer_pool_dump_status;
|
|
break;
|
|
}
|
|
|
|
va_end(ap);
|
|
}
|
|
|
|
/*****************************************************************//**
|
|
Sets the global variable that feeds MySQL's innodb_buffer_pool_load_status
|
|
to the specified string. The format and the following parameters are the
|
|
same as the ones used for printf(3). The value of this variable can be
|
|
retrieved by:
|
|
SELECT variable_value FROM information_schema.global_status WHERE
|
|
variable_name = 'INNODB_BUFFER_POOL_LOAD_STATUS';
|
|
or by:
|
|
SHOW STATUS LIKE 'innodb_buffer_pool_load_status'; */
|
|
static MY_ATTRIBUTE((nonnull, format(printf, 2, 3)))
|
|
void
|
|
buf_load_status(
|
|
/*============*/
|
|
enum status_severity severity,/*!< in: status severity */
|
|
const char* fmt, /*!< in: format */
|
|
...) /*!< in: extra parameters according to fmt */
|
|
{
|
|
va_list ap;
|
|
|
|
va_start(ap, fmt);
|
|
|
|
vsnprintf(
|
|
export_vars.innodb_buffer_pool_load_status,
|
|
sizeof(export_vars.innodb_buffer_pool_load_status),
|
|
fmt, ap);
|
|
|
|
switch (severity) {
|
|
case STATUS_INFO:
|
|
ib::info() << export_vars.innodb_buffer_pool_load_status;
|
|
break;
|
|
|
|
case STATUS_ERR:
|
|
ib::error() << export_vars.innodb_buffer_pool_load_status;
|
|
break;
|
|
}
|
|
|
|
va_end(ap);
|
|
}
|
|
|
|
/** Returns the directory path where the buffer pool dump file will be created.
|
|
@return directory path */
|
|
static
|
|
const char*
|
|
get_buf_dump_dir()
|
|
{
|
|
const char* dump_dir;
|
|
|
|
/* The dump file should be created in the default data directory if
|
|
innodb_data_home_dir is set as an empty string. */
|
|
if (!*srv_data_home) {
|
|
dump_dir = fil_path_to_mysql_datadir;
|
|
} else {
|
|
dump_dir = srv_data_home;
|
|
}
|
|
|
|
return(dump_dir);
|
|
}
|
|
|
|
/** Generate the path to the buffer pool dump/load file.
|
|
@param[out] path generated path
|
|
@param[in] path_size size of 'path', used as in snprintf(3). */
|
|
static void buf_dump_generate_path(char *path, size_t path_size)
|
|
{
|
|
char buf[FN_REFLEN];
|
|
|
|
mysql_mutex_lock(&LOCK_global_system_variables);
|
|
snprintf(buf, sizeof buf, "%s" FN_ROOTDIR "%s", get_buf_dump_dir(),
|
|
srv_buf_dump_filename);
|
|
mysql_mutex_unlock(&LOCK_global_system_variables);
|
|
|
|
os_file_type_t type;
|
|
bool exists = false;
|
|
bool ret;
|
|
|
|
ret = os_file_status(buf, &exists, &type);
|
|
|
|
/* For realpath() to succeed the file must exist. */
|
|
|
|
if (ret && exists) {
|
|
/* my_realpath() assumes the destination buffer is big enough
|
|
to hold FN_REFLEN bytes. */
|
|
ut_a(path_size >= FN_REFLEN);
|
|
|
|
my_realpath(path, buf, 0);
|
|
} else {
|
|
/* If it does not exist, then resolve only srv_data_home
|
|
and append srv_buf_dump_filename to it. */
|
|
char srv_data_home_full[FN_REFLEN];
|
|
|
|
my_realpath(srv_data_home_full, get_buf_dump_dir(), 0);
|
|
const char *format;
|
|
|
|
switch (srv_data_home_full[strlen(srv_data_home_full) - 1]) {
|
|
#ifdef _WIN32
|
|
case '\\':
|
|
#endif
|
|
case '/':
|
|
format = "%s%s";
|
|
break;
|
|
default:
|
|
format = "%s" FN_ROOTDIR "%s";
|
|
}
|
|
|
|
snprintf(path, path_size, format,
|
|
srv_data_home_full, srv_buf_dump_filename);
|
|
}
|
|
}
|
|
|
|
|
|
/*****************************************************************//**
|
|
Perform a buffer pool dump into the file specified by
|
|
innodb_buffer_pool_filename. If any errors occur then the value of
|
|
innodb_buffer_pool_dump_status will be set accordingly, see buf_dump_status().
|
|
The dump filename can be specified by (relative to srv_data_home):
|
|
SET GLOBAL innodb_buffer_pool_filename='filename'; */
|
|
static
|
|
void
|
|
buf_dump(
|
|
/*=====*/
|
|
ibool obey_shutdown) /*!< in: quit if we are in a shutting down
|
|
state */
|
|
{
|
|
#define SHOULD_QUIT() (SHUTTING_DOWN() && obey_shutdown)
|
|
|
|
char full_filename[OS_FILE_MAX_PATH];
|
|
char tmp_filename[OS_FILE_MAX_PATH + sizeof "incomplete"];
|
|
char now[32];
|
|
FILE* f;
|
|
int ret;
|
|
|
|
buf_dump_generate_path(full_filename, sizeof(full_filename));
|
|
|
|
snprintf(tmp_filename, sizeof(tmp_filename),
|
|
"%s.incomplete", full_filename);
|
|
|
|
buf_dump_status(STATUS_INFO, "Dumping buffer pool(s) to %s",
|
|
full_filename);
|
|
|
|
#ifdef _WIN32
|
|
/* use my_fopen() for correct permissions during bootstrap*/
|
|
f = my_fopen(tmp_filename, O_RDWR|O_TRUNC|O_CREAT, 0);
|
|
#elif defined(__GLIBC__) || O_CLOEXEC == 0
|
|
f = fopen(tmp_filename, "w" STR_O_CLOEXEC);
|
|
#else
|
|
{
|
|
int fd;
|
|
fd = open(tmp_filename, O_CREAT | O_TRUNC | O_CLOEXEC | O_WRONLY, 0640);
|
|
if (fd >= 0) {
|
|
f = fdopen(fd, "w");
|
|
}
|
|
else {
|
|
f = NULL;
|
|
}
|
|
}
|
|
#endif
|
|
if (f == NULL) {
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot open '%s' for writing: %s",
|
|
tmp_filename, strerror(errno));
|
|
return;
|
|
}
|
|
const buf_page_t* bpage;
|
|
page_id_t* dump;
|
|
ulint n_pages;
|
|
ulint j;
|
|
|
|
mysql_mutex_lock(&buf_pool.mutex);
|
|
|
|
n_pages = UT_LIST_GET_LEN(buf_pool.LRU);
|
|
|
|
/* skip empty buffer pools */
|
|
if (n_pages == 0) {
|
|
mysql_mutex_unlock(&buf_pool.mutex);
|
|
goto done;
|
|
}
|
|
|
|
if (srv_buf_pool_dump_pct != 100) {
|
|
ulint t_pages;
|
|
|
|
/* limit the number of total pages dumped to X% of the
|
|
total number of pages */
|
|
t_pages = buf_pool.curr_size() * srv_buf_pool_dump_pct / 100;
|
|
if (n_pages > t_pages) {
|
|
buf_dump_status(STATUS_INFO,
|
|
"Restricted to " ULINTPF
|
|
" pages due to "
|
|
"innodb_buf_pool_dump_pct=%lu",
|
|
t_pages, srv_buf_pool_dump_pct);
|
|
n_pages = t_pages;
|
|
}
|
|
|
|
if (n_pages == 0) {
|
|
n_pages = 1;
|
|
}
|
|
}
|
|
|
|
dump = static_cast<page_id_t*>(ut_malloc_nokey(
|
|
n_pages * sizeof(*dump)));
|
|
|
|
if (dump == NULL) {
|
|
std::ostringstream str_bytes;
|
|
mysql_mutex_unlock(&buf_pool.mutex);
|
|
fclose(f);
|
|
str_bytes << ib::bytes_iec{n_pages * sizeof(*dump)};
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot allocate %s: %s",
|
|
str_bytes.str().c_str(),
|
|
strerror(errno));
|
|
/* leave tmp_filename to exist */
|
|
return;
|
|
}
|
|
|
|
for (bpage = UT_LIST_GET_FIRST(buf_pool.LRU), j = 0;
|
|
bpage != NULL && j < n_pages;
|
|
bpage = UT_LIST_GET_NEXT(LRU, bpage)) {
|
|
const auto status = bpage->state();
|
|
if (status < buf_page_t::UNFIXED) {
|
|
ut_a(status >= buf_page_t::FREED);
|
|
continue;
|
|
}
|
|
const page_id_t id{bpage->id()};
|
|
|
|
if (id.space() == SRV_TMP_SPACE_ID) {
|
|
/* Ignore the innodb_temporary tablespace. */
|
|
continue;
|
|
}
|
|
|
|
dump[j++] = id;
|
|
}
|
|
|
|
mysql_mutex_unlock(&buf_pool.mutex);
|
|
|
|
ut_a(j <= n_pages);
|
|
n_pages = j;
|
|
|
|
for (j = 0; j < n_pages && !SHOULD_QUIT(); j++) {
|
|
ret = fprintf(f, "%u,%u\n",
|
|
dump[j].space(), dump[j].page_no());
|
|
if (ret < 0) {
|
|
ut_free(dump);
|
|
fclose(f);
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot write to '%s': %s",
|
|
tmp_filename, strerror(errno));
|
|
/* leave tmp_filename to exist */
|
|
return;
|
|
}
|
|
if (SHUTTING_DOWN() && !(j & 1023)) {
|
|
service_manager_extend_timeout(
|
|
INNODB_EXTEND_TIMEOUT_INTERVAL,
|
|
"Dumping buffer pool page "
|
|
ULINTPF "/" ULINTPF, j + 1, n_pages);
|
|
}
|
|
}
|
|
|
|
ut_free(dump);
|
|
|
|
done:
|
|
ret = IF_WIN(my_fclose(f,0),fclose(f));
|
|
if (ret != 0) {
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot close '%s': %s",
|
|
tmp_filename, strerror(errno));
|
|
return;
|
|
}
|
|
/* else */
|
|
|
|
ret = unlink(full_filename);
|
|
if (ret != 0 && errno != ENOENT) {
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot delete '%s': %s",
|
|
full_filename, strerror(errno));
|
|
/* leave tmp_filename to exist */
|
|
return;
|
|
}
|
|
/* else */
|
|
|
|
ret = rename(tmp_filename, full_filename);
|
|
if (ret != 0) {
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot rename '%s' to '%s': %s",
|
|
tmp_filename, full_filename,
|
|
strerror(errno));
|
|
/* leave tmp_filename to exist */
|
|
return;
|
|
}
|
|
/* else */
|
|
|
|
/* success */
|
|
|
|
ut_sprintf_timestamp(now);
|
|
|
|
buf_dump_status(STATUS_INFO,
|
|
"Buffer pool(s) dump completed at %s", now);
|
|
|
|
/* Though dumping doesn't related to an incomplete load,
|
|
we reset this to 0 here to indicate that a shutdown can also perform
|
|
a dump */
|
|
export_vars.innodb_buffer_pool_load_incomplete = 0;
|
|
}
|
|
|
|
/*****************************************************************//**
|
|
Perform a buffer pool load from the file specified by
|
|
innodb_buffer_pool_filename. If any errors occur then the value of
|
|
innodb_buffer_pool_load_status will be set accordingly, see buf_load_status().
|
|
The dump filename can be specified by (relative to srv_data_home):
|
|
SET GLOBAL innodb_buffer_pool_filename='filename'; */
|
|
static
|
|
void
|
|
buf_load()
|
|
/*======*/
|
|
{
|
|
char full_filename[OS_FILE_MAX_PATH];
|
|
char now[32];
|
|
FILE* f;
|
|
page_id_t* dump;
|
|
ulint dump_n;
|
|
ulint i;
|
|
uint32_t space_id;
|
|
uint32_t page_no;
|
|
int fscanf_ret;
|
|
|
|
/* Ignore any leftovers from before */
|
|
buf_load_abort_flag = false;
|
|
|
|
buf_dump_generate_path(full_filename, sizeof(full_filename));
|
|
|
|
buf_load_status(STATUS_INFO,
|
|
"Loading buffer pool(s) from %s", full_filename);
|
|
|
|
f = fopen(full_filename, "r" STR_O_CLOEXEC);
|
|
if (f == NULL) {
|
|
buf_load_status(STATUS_INFO,
|
|
"Cannot open '%s' for reading: %s",
|
|
full_filename, strerror(errno));
|
|
return;
|
|
}
|
|
/* else */
|
|
|
|
/* First scan the file to estimate how many entries are in it.
|
|
This file is tiny (approx 500KB per 1GB buffer pool), reading it
|
|
two times is fine. */
|
|
dump_n = 0;
|
|
while (fscanf(f, "%u,%u", &space_id, &page_no) == 2
|
|
&& !SHUTTING_DOWN()) {
|
|
dump_n++;
|
|
}
|
|
|
|
if (!SHUTTING_DOWN() && !feof(f)) {
|
|
/* fscanf() returned != 2 */
|
|
const char* what;
|
|
if (ferror(f)) {
|
|
what = "reading";
|
|
} else {
|
|
what = "parsing";
|
|
}
|
|
fclose(f);
|
|
buf_load_status(STATUS_ERR, "Error %s '%s',"
|
|
" unable to load buffer pool (stage 1)",
|
|
what, full_filename);
|
|
return;
|
|
}
|
|
|
|
/* If the dump is larger than the buffer pool, then we ignore the
|
|
extra trailing. This could happen if a dump is made, then buffer
|
|
pool is shrunk and then load is attempted. */
|
|
dump_n = std::min(dump_n, buf_pool.curr_size());
|
|
|
|
if (dump_n != 0) {
|
|
dump = static_cast<page_id_t*>(ut_malloc_nokey(
|
|
dump_n * sizeof(*dump)));
|
|
} else {
|
|
fclose(f);
|
|
ut_sprintf_timestamp(now);
|
|
buf_load_status(STATUS_INFO,
|
|
"Buffer pool(s) load completed at %s"
|
|
" (%s was empty)", now, full_filename);
|
|
return;
|
|
}
|
|
|
|
if (dump == NULL) {
|
|
std::ostringstream str_bytes;
|
|
fclose(f);
|
|
str_bytes << ib::bytes_iec{dump_n * sizeof(*dump)};
|
|
buf_dump_status(STATUS_ERR,
|
|
"Cannot allocate %s: %s",
|
|
str_bytes.str().c_str(),
|
|
strerror(errno));
|
|
/* leave tmp_filename to exist */
|
|
return;
|
|
}
|
|
|
|
rewind(f);
|
|
|
|
export_vars.innodb_buffer_pool_load_incomplete = 1;
|
|
|
|
for (i = 0; i < dump_n && !SHUTTING_DOWN(); i++) {
|
|
fscanf_ret = fscanf(f, "%u,%u", &space_id, &page_no);
|
|
|
|
if (fscanf_ret != 2) {
|
|
if (feof(f)) {
|
|
break;
|
|
}
|
|
/* else */
|
|
|
|
ut_free(dump);
|
|
fclose(f);
|
|
buf_load_status(STATUS_ERR,
|
|
"Error parsing '%s', unable"
|
|
" to load buffer pool (stage 2)",
|
|
full_filename);
|
|
return;
|
|
}
|
|
|
|
if (space_id > ULINT32_MASK || page_no > ULINT32_MASK) {
|
|
ut_free(dump);
|
|
fclose(f);
|
|
buf_load_status(STATUS_ERR,
|
|
"Error parsing '%s': bogus"
|
|
" space,page %u,%u at line " ULINTPF
|
|
", unable to load buffer pool",
|
|
full_filename,
|
|
space_id, page_no,
|
|
i);
|
|
return;
|
|
}
|
|
|
|
dump[i] = page_id_t(space_id, page_no);
|
|
}
|
|
|
|
/* Set dump_n to the actual number of initialized elements,
|
|
i could be smaller than dump_n here if the file got truncated after
|
|
we read it the first time. */
|
|
dump_n = i;
|
|
|
|
fclose(f);
|
|
|
|
if (dump_n == 0) {
|
|
ut_free(dump);
|
|
ut_sprintf_timestamp(now);
|
|
buf_load_status(STATUS_INFO,
|
|
"Buffer pool(s) load completed at %s"
|
|
" (%s was empty or had errors)", now, full_filename);
|
|
return;
|
|
}
|
|
|
|
if (!SHUTTING_DOWN()) {
|
|
std::sort(dump, dump + dump_n);
|
|
std::set<uint32_t> missing;
|
|
for (const page_id_t id : st_::span<const page_id_t>
|
|
(dump, dump_n)) {
|
|
missing.emplace(id.space());
|
|
}
|
|
for (std::set<uint32_t>::iterator i = missing.begin();
|
|
i != missing.end(); ) {
|
|
auto j = i++;
|
|
if (fil_space_t* space = fil_space_t::get(*j)) {
|
|
space->release();
|
|
missing.erase(j);
|
|
}
|
|
}
|
|
if (!missing.empty()) {
|
|
dict_check_tablespaces_and_store_max_id(&missing);
|
|
}
|
|
}
|
|
|
|
/* Avoid calling the expensive fil_space_t::get() for each
|
|
page within the same tablespace. dump[] is sorted by (space, page),
|
|
so all pages from a given tablespace are consecutive. */
|
|
uint32_t cur_space_id = dump[0].space();
|
|
fil_space_t* space = fil_space_t::get(cur_space_id);
|
|
ulint zip_size = space ? space->zip_size() : 0;
|
|
|
|
PSI_stage_progress* pfs_stage_progress __attribute__((unused))
|
|
= mysql_set_stage(srv_stage_buffer_pool_load.m_key);
|
|
mysql_stage_set_work_estimated(pfs_stage_progress, dump_n);
|
|
mysql_stage_set_work_completed(pfs_stage_progress, 0);
|
|
|
|
for (i = 0; i < dump_n && !SHUTTING_DOWN(); i++) {
|
|
|
|
/* space_id for this iteration of the loop */
|
|
const uint32_t this_space_id = dump[i].space();
|
|
|
|
if (this_space_id >= SRV_SPACE_ID_UPPER_BOUND) {
|
|
continue;
|
|
}
|
|
|
|
if (this_space_id != cur_space_id) {
|
|
if (space) {
|
|
space->release();
|
|
}
|
|
|
|
cur_space_id = this_space_id;
|
|
space = fil_space_t::get(cur_space_id);
|
|
|
|
if (!space) {
|
|
continue;
|
|
}
|
|
|
|
zip_size = space->zip_size();
|
|
}
|
|
|
|
/* JAN: TODO: As we use background page read below,
|
|
if tablespace is encrypted we cant use it. */
|
|
if (!space || dump[i].page_no() >= space->get_size() ||
|
|
(space->crypt_data &&
|
|
space->crypt_data->encryption != FIL_ENCRYPTION_OFF &&
|
|
space->crypt_data->type != CRYPT_SCHEME_UNENCRYPTED)) {
|
|
continue;
|
|
}
|
|
|
|
if (space->is_stopping()) {
|
|
space->release();
|
|
space = nullptr;
|
|
continue;
|
|
}
|
|
|
|
space->reacquire();
|
|
buf_read_page_background(space, dump[i], zip_size);
|
|
|
|
if (buf_load_abort_flag) {
|
|
if (space) {
|
|
space->release();
|
|
}
|
|
buf_load_abort_flag = false;
|
|
ut_free(dump);
|
|
buf_load_status(
|
|
STATUS_INFO,
|
|
"Buffer pool(s) load aborted on request");
|
|
/* Premature end, set estimated = completed = i and
|
|
end the current stage event. */
|
|
|
|
mysql_stage_set_work_estimated(pfs_stage_progress, i);
|
|
mysql_stage_set_work_completed(pfs_stage_progress, i);
|
|
|
|
mysql_end_stage();
|
|
return;
|
|
}
|
|
|
|
#ifdef UNIV_DEBUG
|
|
if ((i+1) >= srv_buf_pool_load_pages_abort) {
|
|
buf_load_abort_flag = true;
|
|
}
|
|
#endif
|
|
}
|
|
|
|
if (space) {
|
|
space->release();
|
|
}
|
|
|
|
ut_free(dump);
|
|
|
|
os_aio_wait_until_no_pending_reads(true);
|
|
|
|
ut_sprintf_timestamp(now);
|
|
|
|
if (i == dump_n) {
|
|
buf_load_status(STATUS_INFO,
|
|
"Buffer pool(s) load completed at %s", now);
|
|
export_vars.innodb_buffer_pool_load_incomplete = 0;
|
|
} else if (!buf_load_abort_flag) {
|
|
buf_load_status(STATUS_INFO,
|
|
"Buffer pool(s) load aborted due to user instigated abort at %s",
|
|
now);
|
|
/* intentionally don't reset innodb_buffer_pool_load_incomplete
|
|
as we don't want a shutdown to save the buffer pool */
|
|
} else {
|
|
buf_load_status(STATUS_INFO,
|
|
"Buffer pool(s) load aborted due to shutdown at %s",
|
|
now);
|
|
/* intentionally don't reset innodb_buffer_pool_load_incomplete
|
|
as we want to abort without saving the buffer pool */
|
|
}
|
|
|
|
/* Make sure that estimated = completed when we end. */
|
|
mysql_stage_set_work_completed(pfs_stage_progress, dump_n);
|
|
/* End the stage progress event. */
|
|
mysql_end_stage();
|
|
}
|
|
|
|
/** Abort a currently running buffer pool load. */
|
|
void buf_load_abort()
|
|
{
|
|
buf_load_abort_flag= true;
|
|
}
|
|
|
|
/*****************************************************************//**
|
|
This is the main task for buffer pool dump/load. when scheduled
|
|
either performs a dump or load, depending on server state, state of the variables etc- */
|
|
static void buf_dump_load_func(void *)
|
|
{
|
|
ut_ad(!srv_read_only_mode);
|
|
static bool first_time = true;
|
|
if (first_time && srv_buffer_pool_load_at_startup) {
|
|
|
|
#ifdef WITH_WSREP
|
|
if (!get_wsrep_recovery()) {
|
|
#endif /* WITH_WSREP */
|
|
srv_thread_pool->set_concurrency(srv_n_read_io_threads);
|
|
buf_load();
|
|
srv_thread_pool->set_concurrency();
|
|
#ifdef WITH_WSREP
|
|
}
|
|
#endif /* WITH_WSREP */
|
|
}
|
|
first_time = false;
|
|
|
|
while (!SHUTTING_DOWN()) {
|
|
if (buf_dump_should_start) {
|
|
buf_dump_should_start = false;
|
|
buf_dump(true);
|
|
}
|
|
if (buf_load_should_start) {
|
|
buf_load_should_start = false;
|
|
buf_load();
|
|
}
|
|
|
|
if (!buf_dump_should_start && !buf_load_should_start) {
|
|
return;
|
|
}
|
|
}
|
|
|
|
/* In shutdown */
|
|
if (srv_buffer_pool_dump_at_shutdown && srv_fast_shutdown != 2) {
|
|
if (export_vars.innodb_buffer_pool_load_incomplete) {
|
|
buf_dump_status(STATUS_INFO,
|
|
"Dumping of buffer pool not started"
|
|
" as load was incomplete");
|
|
#ifdef WITH_WSREP
|
|
} else if (get_wsrep_recovery()) {
|
|
#endif /* WITH_WSREP */
|
|
} else {
|
|
buf_dump(false/* do complete dump at shutdown */);
|
|
}
|
|
}
|
|
}
|
|
|
|
|
|
/* Execute task with max.concurrency */
|
|
static tpool::task_group tpool_group(1);
|
|
static tpool::waitable_task buf_dump_load_task(buf_dump_load_func, &tpool_group);
|
|
static bool load_dump_enabled;
|
|
|
|
/** Start async buffer pool load, if srv_buffer_pool_load_at_startup was set.*/
|
|
void buf_load_at_startup()
|
|
{
|
|
load_dump_enabled= true;
|
|
if (srv_buffer_pool_load_at_startup)
|
|
buf_do_load_dump();
|
|
}
|
|
|
|
static void buf_do_load_dump()
|
|
{
|
|
if (load_dump_enabled && !buf_dump_load_task.is_running())
|
|
srv_thread_pool->submit_task(&buf_dump_load_task);
|
|
}
|
|
|
|
/** Wait for currently running load/dumps to finish*/
|
|
void buf_load_dump_end()
|
|
{
|
|
ut_ad(SHUTTING_DOWN());
|
|
buf_dump_load_task.wait();
|
|
}
|