水平有限，有误请谅解。
源码版本5.7.22
一、前文总结
前文
MySQL:MGR 学习(1):写集合(Write set) 中
已经说明了Write set的生成过程，但是Write set是需要封装如下Transaction_context_log_event中进行广播到其他节点进行认证的。本文就描述Write set的写入和广播的过程。
如前文所描述，整个事物的Write set在函数binlog_log_row中生成，对于5.7来讲每一行的每个唯一键都会生成一个Write set(但是咨询宋利兵老师得知8.0唯一键不会再记录Write set了)，每个Write set实际上是一个8字节的uint64类型其通过hash函数生成，并且在Rpl_transaction_write_set_ctx存储了一个vector数组和一个set集合来分别存储，如果修改的行比较多那么可能需要一个更多内存来存储这些hash值，虽然8字节比较小，但是如果是大事物上千万的表在一个事物里面做修改那么内存可能消耗会上百兆。如下图是事物执行期间(commit之前)最终形成的Write set内存空间示意图。
二、Transaction_context_log_event生成的时机
在事物执行期间会生成map event/query event/dml event等并且会源源不断的写入到binlog cache中，同时会将Write set 不断的写入到Rpl_transaction_write_set_ctx保存在内存中，这些逻辑都在binlog_log_row中。但是Transaction_context_log_event的生成却是在commit的时候，具体的位置是在MYSQL_BIN_LOG::prepare之后但是在MYSQL_BIN_LOG::ordered_commit之前，显而易见这个时候的binlog event还在bing cache中，还没有写入binlog file中。所以MGR的事物全局认证的动作是发生在binlog event落地之前。下面是这个栈帧：
#0 group_replication_trans_before_commit (param=0x7ffff0e7b8d0) at /root/softm/percona-server-5.7.22-22/rapid/plugin/group_replication/src/observer_trans.cc:511
#1 0x00000000014e4814 in Trans_delegate::before_commit (this=0x2e44800, thd=0x7fffd8000df0, all=false, trx_cache_log=0x7fffd8907a10, stmt_cache_log=0x7fffd8907858,
cache_log_max_size=18446744073709547520) at /root/softm/percona-server-5.7.22-22/sql/rpl_handler.cc:325
#2 0x000000000188a386 in MYSQL_BIN_LOG::commit (this=0x2e7b440, thd=0x7fffd8000df0, all=false) at /root/softm/percona-server-5.7.22-22/sql/binlog.cc:8974
#3 0x0000000000f80623 in ha_commit_trans (thd=0x7fffd8000df0, all=false, ignore_global_read_lock=false) at /root/softm/percona-server-5.7.22-22/sql/handler.cc:1830
#4 0x00000000016ddab9 in trans_commit_stmt (thd=0x7fffd8000df0) at /root/softm/percona-server-5.7.22-22/sql/transaction.cc:458
#5 0x00000000015d1a8d in mysql_execute_command (thd=0x7fffd8000df0, first_level=true) at /root/softm/percona-server-5.7.22-22/sql/sql_parse.cc:5293
#6 0x00000000015d3182 in mysql_parse (thd=0x7fffd8000df0, parser_state=0x7ffff0e7e600) at /root/softm/percona-server-5.7.22-22/sql/sql_parse.cc:5901
#7 0x00000000015c6d16 in dispatch_command (thd=0x7fffd8000df0, com_data=0x7ffff0e7ed70, command=COM_QUERY)
at /root/softm/percona-server-5.7.22-22/sql/sql_parse.cc:1490
#8 0x00000000015c5aa3 in do_command (thd=0x7fffd8000df0) at /root/softm/percona-server-5.7.22-22/sql/sql_parse.cc:1021
#9 0x000000000170ebb0 in handle_connection (arg=0x3cd32d0) at /root/softm/percona-server-5.7.22-22/sql/conn_handler/connection_handler_per_thread.cc:312
#10 0x0000000001946140 in pfs_spawn_thread (arg=0x3c71630) at /root/softm/percona-server-5.7.22-22/storage/perfschema/pfs.cc:2190
#11 0x00007ffff7bc7851 in start_thread () from /lib64/libpthread.so.0
#12 0x00007ffff651290d in clone () from /lib64/libc.so.6三、MGR全局认证发送内容的生成过程
下面是我通过对源码浅显的理解得出过程：
- 1、获取当前的binlog cache内容记录为cache_log，这些就是已经在执行阶段生成map/query/dml event等。
- 2、生成一个新的IO_CACHE作为临时存储为cache，目的在于存储。Transaction_context_log_event 和Gtid_log_event。
- 3、将cache_log类型转换为READ类型同时初始化各种辅助类容如偏移量。
- 4、初始化Transaction_context_log_event 。
- 5、扫描Rpl_transaction_write_set_ctx中的write_set_unique 集合的内容，并且将其存储到Transaction_write_set 定义的内存空间中write_set中，注意这里只是用到了集合没用到数组。这里也就是进行Write set的一个拷贝而已其考到write_set临时变量中。
- 6、将write_set内容填充到Transaction_context_log_event中，整个过程还会做base64的转换，最终填充到event的是base64格式的Write set类容。完成后析构write_set来临时变量
- 7、 将Transaction_context_log_event写入到第二步定义的cache中。
- 8、生成Gtid_log_event，只是做一些初始化动作，Gtid并没有生成。
- 9、将Gtid_log_event写入到第二步定义的cache中。
- 10、通过cache+cache_log的总和来对比group_replication_transaction_size_limit设置的值，也就是判断整个事物的binlog event是否操作了参数设置。
- 11、将cache类型转换为READ类型同时初始化各种辅助类容如偏移量。
- 12、将cache和cache_log分别写入到到transaction_msg中。
- 13、流控相关，没仔细看，如果有机会学习流控机制在仔细学习。
- 14、gcs_module负责发送transaction_msg到各个节点
- 15、挂起等待事物认证的结果。
那么整个过程大概就是：
- 经过hash的Write set (集合)->拷贝到write_set变量(类数组)->通过base64算法写入到Transaction_context_log_event ->合并其他binlog event到transaction_msg->gcs_module广播transaction_msg到其他节点->等待认证结果
四、相关源码
1、group_replication_trans_before_commit 函数相关内容
if (trx_cache_log_position > 0 && stmt_cache_log_position == 0) //如果存在事物cache
{
cache_log= param->trx_cache_log; //设置到IO_cache
cache_log_position= trx_cache_log_position;
}
else if (trx_cache_log_position == 0 && stmt_cache_log_position > 0)//如果存在语句cache
{
cache_log= param->stmt_cache_log;
cache_log_position= stmt_cache_log_position;
is_dml= false;
may_have_sbr_stmts= true;
}
else
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "We can only use one cache type at a "
"time on session %u", param->thread_id);
shared_plugin_stop_lock->release_read_lock();
DBUG_RETURN(1);
/* purecov: end */
}
applier_module->get_pipeline_stats_member_collector()
->increment_transactions_local();
DBUG_ASSERT(cache_log->type == WRITE_CACHE);
DBUG_PRINT("cache_log", ("thread_id: %u, trx_cache_log_position: %llu,"
" stmt_cache_log_position: %llu",
param->thread_id, trx_cache_log_position,
stmt_cache_log_position));
/*
Open group replication cache.
Reuse the same cache on each session for improved performance.
*/
cache= observer_trans_get_io_cache(param->thread_id,
param->cache_log_max_size); //获取一个新的IO_CACHE系统
if (cache == NULL) //错误处理
{
/* purecov: begin inspected */
error= pre_wait_error;
goto err;
/* purecov: end */
}
// Reinit binlog cache to read.
if (reinit_cache(cache_log, READ_CACHE, 0)) ////将IO_CACHE类型进行转换 并且位置还原
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Failed to reinit binlog cache log for read "
"on session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
/*
After this, cache_log should be reinit to old saved value when we
are going out of the function scope.
*/
reinit_cache_log_required= true;
// Create transaction context.
tcle= new Transaction_context_log_event(param->server_uuid, Rpl_transaction_write_set_ctx
is_dml,
param->thread_id,
is_gtid_specified); //初始化 Transaction_context_log_event
if (!tcle->is_valid())
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL,
"Failed to create the context of the current "
"transaction on session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
if (is_dml)
{
Transaction_write_set* write_set= get_transaction_write_set(param->thread_id);// 获取前期得到write set 并且放回到一个临时内存空间 write_set中
/*
When GTID is specified we may have empty transactions, that is,
a transaction may have not write set at all because it didn't
change any data, it will just persist that GTID as applied.
*/
if ((write_set == NULL) && (!is_gtid_specified))
{
log_message(MY_ERROR_LEVEL, "Failed to extract the set of items written "
"during the execution of the current "
"transaction on session %u", param->thread_id);
error= pre_wait_error;
goto err;
}
if (write_set != NULL)
{
if (add_write_set(tcle, write_set))//将整个wirte_set内容复制到event Transaction_context_log_event中 此时就进入了event了
{
/* purecov: begin inspected */
cleanup_transaction_write_set(write_set); //write set已经完成了它的功能需要析构
log_message(MY_ERROR_LEVEL, "Failed to gather the set of items written "
"during the execution of the current "
"transaction on session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
cleanup_transaction_write_set(write_set); //如果add_write_set函数调用出现 有问题 也需要析构掉
DBUG_ASSERT(is_gtid_specified || (tcle->get_write_set()->size() > 0));
}
else
{
/*
For empty transactions we should set the GTID may_have_sbr_stmts. See
comment at binlog_cache_data::may_have_sbr_stmts().
*/
may_have_sbr_stmts= true;
} Log_event::write
}
// Write transaction context to group replication cache.
tcle->write(cache); //写入到MGR CACHE 写入 TCLE的header(virtual) body(virtual) footer
// Write Gtid log event to group replication cache.
gle= new Gtid_log_event(param->server_id, is_dml, 0, 1,
may_have_sbr_stmts,
gtid_specification);
gle->write(cache); //写入GTID event到MGR CACHE 占位
transaction_size= cache_log_position + my_b_tell(cache);
if (is_dml && transaction_size_limit &&
transaction_size > transaction_size_limit)
{
log_message(MY_ERROR_LEVEL, "Error on session %u. "
"Transaction of size %llu exceeds specified limit %lu. "
"To increase the limit please adjust group_replication_transaction_size_limit option.",
param->thread_id, transaction_size,
transaction_size_limit); //group_replication_transaction_size_limit 事物大小参数
error= pre_wait_error;
goto err;
}
// Reinit group replication cache to read.
if (reinit_cache(cache, READ_CACHE, 0))//将IO_CACHE类型进行转换 并且位置还原
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Error while re-initializing an internal "
"cache, for read operations, on session %u",
param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
// Copy group replication cache to buffer.
if (transaction_msg.append_cache(cache)) //加入到transaction_msg
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Error while appending data to an internal "
"cache on session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
// Copy binlog cache content to buffer.
if (transaction_msg.append_cache(cache_log))//加入到transaction_msg
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Error while writing binary log cache on "
"session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
DBUG_ASSERT(certification_latch != NULL);
if (certification_latch->registerTicket(param->thread_id))
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Unable to register for getting notifications "
"regarding the outcome of the transaction on "
"session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
#ifndef DBUG_OFF
DBUG_EXECUTE_IF("test_basic_CRUD_operations_sql_service_interface",
{
DBUG_SET("-d,test_basic_CRUD_operations_sql_service_interface");
DBUG_ASSERT(!sql_command_check());
};);
DBUG_EXECUTE_IF("group_replication_before_message_broadcast",
{
const char act[]= "now wait_for waiting";
DBUG_ASSERT(!debug_sync_set_action(current_thd, STRING_WITH_LEN(act)));
});
#endif
/*
Check if member needs to throttle its transactions to avoid
cause starvation on the group.
*/
applier_module->get_flow_control_module()->do_wait(); //流控相关
//Broadcast the Transaction Message
send_error= gcs_module->send_message(transaction_msg); //gcs广播
if (send_error == GCS_MESSAGE_TOO_BIG)
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Error broadcasting transaction to the group "
"on session %u. Message is too big.",
param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
else if (send_error == GCS_NOK)
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Error while broadcasting the transaction to "
"the group on session %u", param->thread_id);
error= pre_wait_error;
goto err;
/* purecov: end */
}
shared_plugin_stop_lock->release_read_lock();
DBUG_ASSERT(certification_latch != NULL);
if (certification_latch->waitTicket(param->thread_id)) //等待认证结果
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "Error while waiting for conflict detection "
"procedure to finish on session %u",
param->thread_id);
error= post_wait_error;
goto err;
/* purecov: end */
}
2、add_write_set函数
int add_write_set(Transaction_context_log_event *tcle,
Transaction_write_set *set)
{
DBUG_ENTER("add_write_set");
int iterator= set->write_set_size; //将循环次数设置为 set的长度 也就是有多少个write sets
for (int i = 0; i < iterator; i++)
{
uchar buff[BUFFER_READ_PKE];
int8store(buff, set->write_set[i]); //逐字节复制到buffer中
uint64 const tmp_str_sz= base64_needed_encoded_length((uint64) BUFFER_READ_PKE);
char *write_set_value= (char *) my_malloc(PSI_NOT_INSTRUMENTED,
static_cast<size_t>(tmp_str_sz), MYF(MY_WME)); //13bytes (gdb) p tmp_str_sz $2 = 13
if (!write_set_value)//分配内存错误
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL, "No memory to generate write identification hash");
DBUG_RETURN(1);
/* purecov: end */
}
if (base64_encode(buff, (size_t) BUFFER_READ_PKE, write_set_value)) //做base64算法
{
/* purecov: begin inspected */
log_message(MY_ERROR_LEVEL,
"Base 64 encoding of the write identification hash failed");
DBUG_RETURN(1);
/* purecov: end */
}
tcle->add_write_set(write_set_value); //最终将base64格式的write set写入到event中
}
DBUG_RETURN(0);
}3、get_transaction_write_set函数
Transaction_write_set* get_transaction_write_set(unsigned long m_thread_id)
{
DBUG_ENTER("get_transaction_write_set");
THD *thd= NULL;
Transaction_write_set *result_set= NULL;
Find_thd_with_id find_thd_with_id(m_thread_id, false);
thd= Global_THD_manager::get_instance()->find_thd(&find_thd_with_id);
if (thd)
{
std::set<uint64> *write_set= thd->get_transaction()
->get_transaction_write_set_ctx()->get_write_set(); //Rpl_transaction_write_set_ctx std::set<uint64> *get_write_set();
unsigned long write_set_size= write_set->size(); //返回集合大小
if (write_set_size == 0)
{
mysql_mutex_unlock(&thd->LOCK_thd_data);
DBUG_RETURN(NULL);
}
result_set= (Transaction_write_set*)my_malloc(key_memory_write_set_extraction,
sizeof(Transaction_write_set),
MYF(0));//这里为其Transaction_write_set分配内存空间
result_set->write_set_size= write_set_size; //获取size
result_set->write_set=
(unsigned long long*)my_malloc(key_memory_write_set_extraction,
write_set_size *
sizeof(unsigned long long),
MYF(0));//分配内存
int result_set_index= 0;
for (std::set<uint64>::iterator it= write_set->begin();//完成复制注意是从set中复制到简单的内存中
it != write_set->end();
++it)
{
uint64 temp= *it;
result_set->write_set[result_set_index++]=temp;
}
mysql_mutex_unlock(&thd->LOCK_thd_data);
}
DBUG_RETURN(result_set);
}
作者微信：
