mirror/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-01-16 12:02:42 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions
mariadb/sql/threadpool.h

167 lines
4.7 KiB
C
Raw Normal View History

Merge 10.4 into 10.5
2020-03-30 17:52:17 +02:00
/* Copyright (C) 2012, 2020, MariaDB
Added copiright, some more comments
2012-02-17 23:27:15 +01:00
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
Update FSF address This commit is based on the work of Michal Schorm, rebased on the earliest MariaDB version. Th command line used to generate this diff was: find ./ -type f \ -exec sed -i -e 's/Foundation, Inc., 59 Temple Place, Suite 330, Boston, /Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, /g' {} \; \ -exec sed -i -e 's/Foundation, Inc. 59 Temple Place.* Suite 330, Boston, /Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, /g' {} \; \ -exec sed -i -e 's/MA.*.....-1307.*USA/MA 02110-1335 USA/g' {} \; \ -exec sed -i -e 's/Foundation, Inc., 59 Temple/Foundation, Inc., 51 Franklin/g' {} \; \ -exec sed -i -e 's/Place, Suite 330, Boston, MA.*02111-1307.*USA/Street, Fifth Floor, Boston, MA 02110-1335 USA/g' {} \; \ -exec sed -i -e 's/MA.*.....-1307/MA 02110-1335/g' {} \;
2019-05-10 19:49:46 +02:00
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1335 USA */
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
Merge 10.4 into 10.5
2020-03-30 17:52:17 +02:00
#pragma once
#ifdef HAVE_POOL_OF_THREADS
MDEV-5205 - MariaDB does not start if more than 128 cpu's are available - thread_pool_size command line option upper limit increased to 100 000 (same as for max_connections) - thread_pool_size system variable upper limit is maximum of 128 or the value given at command line - thread groups are now allocated dynamically Different limit for command line option and system variable was done to avoid additional mutex for all_groups and threadpool_max_size.
2013-11-05 06:18:59 +01:00
#define MAX_THREAD_GROUPS 100000
allow changing thread_pool_size without server restart
2011-12-19 13:28:30 +01:00
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
/* Threadpool parameters */
extern uint threadpool_min_threads; /* Minimum threads in pool */
extern uint threadpool_idle_timeout; /* Shutdown idle worker threads after this timeout */
extern uint threadpool_size; /* Number of parallel executing threads */
MDEV-5205 - MariaDB does not start if more than 128 cpu's are available - thread_pool_size command line option upper limit increased to 100 000 (same as for max_connections) - thread_pool_size system variable upper limit is maximum of 128 or the value given at command line - thread groups are now allocated dynamically Different limit for command line option and system variable was done to avoid additional mutex for all_groups and threadpool_max_size.
2013-11-05 06:18:59 +01:00
extern uint threadpool_max_size;
MDEV-19312 Make throttling interval depend on thread_pool_stall_limit A thread_pool_stall_limit which is smaller than default would result in quicker creation of threads.
2019-05-31 10:28:22 +02:00
extern uint threadpool_stall_limit; /* time interval in milliseconds for stall checks*/
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
extern uint threadpool_max_threads; /* Maximum threads in pool */
Allow for faster creation of threads in corner cases where pool would be overloaded with long non-yielding queries. To allow it, change minimum of thread_pool_stall_limit to be 10 milliseconds. Also introduce a new parameter to oversubscribe a group . Number of threads running in parallel would be higher than it normally should, leading to thrashing, but it may improving preemptiveness, which is useful for the described corner case.
2011-12-31 05:24:11 +01:00
extern uint threadpool_oversubscribe; /* Maximum active threads in group */
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
extern uint threadpool_prio_kickup_timer; /* Time before low prio item gets prio boost */
Add some variables to the generic threadpool, that could help to analyze stalls etc better. - thread_pool_exact_stats - uses high precision timestamp for the time when connection was added to the queue. This timestamp helps calculating queuing time shown in I_S.THREADPOOL_QUEUES entries. - If thread_pool_dedicated_listener is on, then each group will have its own dedicated listener, that does not convert to worker. With this variable on, the queueing time in I_S.THREADPOOL_QUEUES , and actual queue size in I_S.THREADPOOOL_GROUPS will be more exact, since IO request are immediately dequeued from poll, without delay. Part of MDEV-19313.
2019-05-26 13:35:07 +02:00
extern my_bool threadpool_exact_stats; /* Better queueing time stats for information_schema, at small performance cost */
extern my_bool threadpool_dedicated_listener; /* Listener thread does not pick up work items. */
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
#ifdef _WIN32
extern uint threadpool_mode; /* Thread pool implementation , windows or generic */
#define TP_MODE_WINDOWS 0
#define TP_MODE_GENERIC 1
#endif
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
MDEV-19312 Make throttling interval depend on thread_pool_stall_limit A thread_pool_stall_limit which is smaller than default would result in quicker creation of threads.
2019-05-31 10:28:22 +02:00
#define DEFAULT_THREADPOOL_STALL_LIMIT 500U
Threadpool -address review comments
2012-01-15 11:17:45 +01:00
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
struct TP_connection;
MDEV-22990 Threadpool : Optimize network/named pipe IO for Windows This patch reduces the overhead of system calls prior to a query, for threadpool. Previously, 3 system calls were done 1. WSARecv() to get notification of input data from client, asynchronous equivalent of select() in one-thread-per-connection 2. recv(4 bytes) - reading packet header length 3. recv(packet payload) Now there will be usually, just WSARecv(), which pre-reads user data into a buffer, so we spared 2 syscalls Profiler shows the most expensive call WSARecv(16%CPU) becomes 4% CPU, after the patch, benchmark results (network heavy ones like point-select) improve by ~20% The buffer management was rather carefully done to keep buffers together, as Windows would keeps the pages pinned in memory for the duration of async calls. At most 1MB memory is used for the buffers, and overhead per-connection is only 256 bytes, which should cover most of the uses. SSL does not yet use the optmization, so far it does not properly use VIO for reads and writes. Neither one-thread-per-connection would get any benefit, but that should be fine, it is not even default on Windows.
2020-06-26 14:43:56 +02:00
struct st_vio;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
extern void tp_callback(TP_connection *c);
extern void tp_timeout_handler(TP_connection *c);
small cleanups
2012-01-24 03:23:14 +01:00
Get rid of idle thread counter atomic variable. Instead, use function that loops over groups and calculates idle threads for "show status".
2012-01-15 15:41:25 +01:00
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
/*
Threadpool statistics
*/
struct TP_STATISTICS
{
/* Current number of worker thread. */
num_worker_threads my_atomic to Atomic_counter
2020-03-26 23:24:49 +01:00
Atomic_counter<uint32_t> num_worker_threads;
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
};
extern TP_STATISTICS tp_stats;
/* Functions to set threadpool parameters */
extern void tp_set_min_threads(uint val);
extern void tp_set_max_threads(uint val);
address second round review comments
2012-02-16 16:59:04 +01:00
extern void tp_set_threadpool_size(uint val);
Make threadpool_stall_limit variable really dynamic
2011-12-29 21:11:06 +01:00
extern void tp_set_threadpool_stall_limit(uint val);
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
extern int tp_get_idle_thread_count();
extern int tp_get_thread_count();
Initial threadpool implementation for MariaDB 5.5
2011-12-08 19:17:49 +01:00
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
enum TP_PRIORITY {
TP_PRIORITY_HIGH,
TP_PRIORITY_LOW,
TP_PRIORITY_AUTO
};
enum TP_STATE
{
TP_STATE_IDLE,
TP_STATE_RUNNING,
MDEV-21101 unexpected wait_timeout with pool-of-threads Due to restricted size of the threadpool, execution of client queries can be delayed (queued) for a while. This delay was interpreted as client inactivity, and connection is closed, if client idle time + queue time exceeds wait_timeout. But users did not expect queue time to be included into wait_timeout. This patch changes the behavior. We don't close connection anymore, if there is some unread data present on connection, even if wait_timeout is exceeded. Unread data means that client was not idle, it sent a query, which we did not have time to process yet.
2020-07-28 15:59:38 +02:00
TP_STATE_PENDING
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
};
/*
Connection structure, encapsulates THD + structures for asynchronous
IO and pool.
Platform specific parts are specified in subclasses called connection_t,
inside threadpool_win.cc and threadpool_unix.cc
*/
MDEV-19313 Threadpool : provide information schema tables for internals of generic threadpool Added thread_pool_groups, thread_pool_queues, thread_pool_waits and thread_pool_stats tables to information_schema.
2019-05-26 13:25:12 +02:00
class CONNECT;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
struct TP_connection
{
THD* thd;
CONNECT* connect;
TP_STATE state;
TP_PRIORITY priority;
TP_connection(CONNECT *c) :
thd(0),
connect(c),
state(TP_STATE_IDLE),
priority(TP_PRIORITY_HIGH)
{}
Apply clang-tidy to remove empty constructors / destructors This patch is the result of running run-clang-tidy -fix -header-filter=.* -checks='-*,modernize-use-equals-default' . Code style changes have been done on top. The result of this change leads to the following improvements: 1. Binary size reduction. * For a -DBUILD_CONFIG=mysql_release build, the binary size is reduced by ~400kb. * A raw -DCMAKE_BUILD_TYPE=Release reduces the binary size by ~1.4kb. 2. Compiler can better understand the intent of the code, thus it leads to more optimization possibilities. Additionally it enabled detecting unused variables that had an empty default constructor but not marked so explicitly. Particular change required following this patch in sql/opt_range.cc result_keys, an unused template class Bitmap now correctly issues unused variable warnings. Setting Bitmap template class constructor to default allows the compiler to identify that there are no side-effects when instantiating the class. Previously the compiler could not issue the warning as it assumed Bitmap class (being a template) would not be performing a NO-OP for its default constructor. This prevented the "unused variable warning".
2023-02-07 12:57:20 +01:00
virtual ~TP_connection() = default;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
/* Initialize io structures windows threadpool, epoll etc */
virtual int init() = 0;
virtual void set_io_timeout(int sec) = 0;
/* Read for the next client command (async) with specified timeout */
virtual int start_io() = 0;
virtual void wait_begin(int type)= 0;
virtual void wait_end() = 0;
MDEV-22990 Threadpool : Optimize network/named pipe IO for Windows This patch reduces the overhead of system calls prior to a query, for threadpool. Previously, 3 system calls were done 1. WSARecv() to get notification of input data from client, asynchronous equivalent of select() in one-thread-per-connection 2. recv(4 bytes) - reading packet header length 3. recv(packet payload) Now there will be usually, just WSARecv(), which pre-reads user data into a buffer, so we spared 2 syscalls Profiler shows the most expensive call WSARecv(16%CPU) becomes 4% CPU, after the patch, benchmark results (network heavy ones like point-select) improve by ~20% The buffer management was rather carefully done to keep buffers together, as Windows would keeps the pages pinned in memory for the duration of async calls. At most 1MB memory is used for the buffers, and overhead per-connection is only 256 bytes, which should cover most of the uses. SSL does not yet use the optmization, so far it does not properly use VIO for reads and writes. Neither one-thread-per-connection would get any benefit, but that should be fine, it is not even default on Windows.
2020-06-26 14:43:56 +02:00
IF_WIN(virtual,) void init_vio(st_vio *){};
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
};
struct TP_pool
{
Apply clang-tidy to remove empty constructors / destructors This patch is the result of running run-clang-tidy -fix -header-filter=.* -checks='-*,modernize-use-equals-default' . Code style changes have been done on top. The result of this change leads to the following improvements: 1. Binary size reduction. * For a -DBUILD_CONFIG=mysql_release build, the binary size is reduced by ~400kb. * A raw -DCMAKE_BUILD_TYPE=Release reduces the binary size by ~1.4kb. 2. Compiler can better understand the intent of the code, thus it leads to more optimization possibilities. Additionally it enabled detecting unused variables that had an empty default constructor but not marked so explicitly. Particular change required following this patch in sql/opt_range.cc result_keys, an unused template class Bitmap now correctly issues unused variable warnings. Setting Bitmap template class constructor to default allows the compiler to identify that there are no side-effects when instantiating the class. Previously the compiler could not issue the warning as it assumed Bitmap class (being a template) would not be performing a NO-OP for its default constructor. This prevented the "unused variable warning".
2023-02-07 12:57:20 +01:00
virtual ~TP_pool() = default;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
virtual int init()= 0;
virtual TP_connection *new_connection(CONNECT *)= 0;
virtual void add(TP_connection *c)= 0;
virtual int set_max_threads(uint){ return 0; }
virtual int set_min_threads(uint){ return 0; }
virtual int set_pool_size(uint){ return 0; }
virtual int set_idle_timeout(uint){ return 0; }
virtual int set_oversubscribe(uint){ return 0; }
virtual int set_stall_limit(uint){ return 0; }
virtual int get_thread_count() { return tp_stats.num_worker_threads; }
virtual int get_idle_thread_count(){ return 0; }
MDEV-24341 Innodb - do not block in foreground thread in log_write_up_to(
2021-02-14 18:30:39 +01:00
virtual void resume(TP_connection* c)=0;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
};
#ifdef _WIN32
MDEV-22990 Threadpool : Optimize network/named pipe IO for Windows This patch reduces the overhead of system calls prior to a query, for threadpool. Previously, 3 system calls were done 1. WSARecv() to get notification of input data from client, asynchronous equivalent of select() in one-thread-per-connection 2. recv(4 bytes) - reading packet header length 3. recv(packet payload) Now there will be usually, just WSARecv(), which pre-reads user data into a buffer, so we spared 2 syscalls Profiler shows the most expensive call WSARecv(16%CPU) becomes 4% CPU, after the patch, benchmark results (network heavy ones like point-select) improve by ~20% The buffer management was rather carefully done to keep buffers together, as Windows would keeps the pages pinned in memory for the duration of async calls. At most 1MB memory is used for the buffers, and overhead per-connection is only 256 bytes, which should cover most of the uses. SSL does not yet use the optmization, so far it does not properly use VIO for reads and writes. Neither one-thread-per-connection would get any benefit, but that should be fine, it is not even default on Windows.
2020-06-26 14:43:56 +02:00
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
struct TP_pool_win:TP_pool
{
Remove some trailing whitespaces.
2020-05-29 12:21:27 +02:00
TP_pool_win();
MDEV-33746 Supply missing override markings Find and fix missing virtual override markings. Updates cmake maintainer flags to include -Wsuggest-override and -Winconsistent-missing-override.
2024-06-12 15:46:26 +02:00
int init() override;
~TP_pool_win() override;
TP_connection *new_connection(CONNECT *c) override;
void add(TP_connection *) override;
int set_max_threads(uint) override;
int set_min_threads(uint) override;
Merge 10.5 into 10.6
2024-06-24 12:09:47 +02:00
void resume(TP_connection *c) override;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
};
#endif
struct TP_pool_generic :TP_pool
{
TP_pool_generic();
~TP_pool_generic();
MDEV-33746 Supply missing override markings Find and fix missing virtual override markings. Updates cmake maintainer flags to include -Wsuggest-override and -Winconsistent-missing-override.
2024-06-12 15:46:26 +02:00
int init() override;
TP_connection *new_connection(CONNECT *c) override;
void add(TP_connection *) override;
int set_pool_size(uint) override;
int set_stall_limit(uint) override;
int get_idle_thread_count() override;
Merge 10.5 into 10.6
2024-06-24 12:09:47 +02:00
void resume(TP_connection* c) override;
MDEV-10297 Add priorization to threadpool Also MDEV-10385 Threadpool refactoring
2016-09-21 16:28:42 +02:00
};
MDEV-19274 mariadb does not build on OSes that do not have HAVE_POOL_OF_THREADS Do not try to compile threadpool_common.cc, if threadpool is not available on current OS. Also introduced undocumented/uncached CMake variable DISABLE_THREADPOOL, to emulate builds on exotic OSes.
2019-04-18 06:50:59 +02:00
#endif /* HAVE_POOL_OF_THREADS */
Reference in a new issue Copy permalink