#At file:///home/andrei/MySQL/BZR/2a-23May/WL/mysql-next-mr-wl5569/ based on revid:andrei.elkin@stripped
3284 Andrei Elkin 2011-06-10
wl#5569 MTS
This patch implements Seconds_behind_master (SBM) updating policy for MTS.
The status parameter is updated per some number of processed group of events.
The number can't be greater than @@global.slave_checkpoint_group and anyway SBM updating
rate does not exceed @@global.slave_checkpoint_period.
Notice that SBM is set to a new value after processing the terminal event (e.g Commit) of a group.
Coordinator resets SBM when notices no more groups left neither to read from Relay-log
nor to process by Workers.
@ mysql-test/suite/rpl/t/rpl_parallel_seconds_behind_master.test
SBM policy for MTS tests are added.
@ mysql-test/suite/rpl/t/rpl_seconds_behind_master.test
the currrent version of the test is MTS unfriendly.
@ sql/log_event.cc
Deploying association of the being assigned group with its terminal event's
timestamp corresponding to end of the event execution.
Making Rotate's update_pos method to modify SBM explicitly through
reset_notified_checkpoint() rathen than mts_checkpoint_routine().
@ sql/rpl_rli.cc
Making reset_notified_checkpoint() to install a new value of SBM.
@ sql/rpl_rli.h
Making reset_notified_checkpoint() to install a new value of SBM.
@ sql/rpl_rli_pdb.h
|Adding new member to transfer SBM of the group to Coordinator which
it installs whenever it processes GAQ effectively.
@ sql/rpl_slave.cc
Coordinator installs SBM as timestamp of the last group discared from GAQ.
Wait for new event Coordinator's read-loop is augmented to implement
SBM resetting.
added:
mysql-test/suite/rpl/r/rpl_parallel_seconds_behind_master.result
mysql-test/suite/rpl/t/rpl_parallel_seconds_behind_master.test
modified:
mysql-test/suite/rpl/t/rpl_seconds_behind_master.test
sql/log_event.cc
sql/rpl_rli.cc
sql/rpl_rli.h
sql/rpl_rli_pdb.h
sql/rpl_slave.cc
=== added file 'mysql-test/suite/rpl/r/rpl_parallel_seconds_behind_master.result'
--- a/mysql-test/suite/rpl/r/rpl_parallel_seconds_behind_master.result 1970-01-01 00:00:00 +0000
+++ b/mysql-test/suite/rpl/r/rpl_parallel_seconds_behind_master.result 2011-06-10 08:04:00 +0000
@@ -0,0 +1,35 @@
+include/master-slave.inc
+[connection master]
+call mtr.add_suppression('Unsafe statement written to the binary log using statement format since BINLOG_FORMAT = STATEMENT');
+include/stop_slave.inc
+set @save.mts_slave_parallel_workers= @@global.mts_slave_parallel_workers;
+set @@global.mts_slave_parallel_workers= 4;
+set @save.mts_checkpoint_period= @@global.mts_checkpoint_period;
+set @@global.mts_checkpoint_period= 500;
+set @save.slave_transaction_retries= @@global.slave_transaction_retries;
+start slave io_thread;
+include/wait_for_slave_io_to_start.inc
+create table t1 (f1 int) engine=innodb;
+create table t2 (f1 int) engine=innodb;
+begin;
+insert into t1 values (sleep(3) + 1);
+Warnings:
+Note 1592 Unsafe statement written to the binary log using statement format since BINLOG_FORMAT = STATEMENT. Statement is unsafe because it uses a system function that may return a different value on the slave.
+commit;
+begin;
+insert into t2 values (1);
+commit;
+include/start_slave.inc
+include/sync_slave_io_with_master.inc
+lock table t2 write;
+First transaction is done, now get Seconds_Behind_Master after it...
+include/wait_for_slave_param.inc [Seconds_Behind_Master]
+Seconds_Behind_Master after first transaction is as expected.
+unlock tables;
+include/wait_for_slave_param.inc [Seconds_Behind_Master]
+drop tables t1, t2;
+set @@global.mts_checkpoint_period= @save.mts_checkpoint_period;
+set @@global.mts_slave_parallel_workers= @save.mts_slave_parallel_workers;
+set @@global.slave_transaction_retries= @save.slave_transaction_retries;
+include/rpl_end.inc
+end of the tests
=== added file 'mysql-test/suite/rpl/t/rpl_parallel_seconds_behind_master.test'
--- a/mysql-test/suite/rpl/t/rpl_parallel_seconds_behind_master.test 1970-01-01 00:00:00 +0000
+++ b/mysql-test/suite/rpl/t/rpl_parallel_seconds_behind_master.test 2011-06-10 08:04:00 +0000
@@ -0,0 +1,141 @@
+#
+# Testing of Seconds_Behind_Master (SBM) to behave as the following
+# SBM changes discretely per some number of processed group of events,
+# e.g transactions.
+# The number is either @@global.slave_checkpoint_group or less if
+# @@global.slave_checkpoint_period timer elapses first.
+# The value updates *after* the last group commit is executed.
+# Resetting to zero behaviour when Slave goes to read events is
+# preserved.
+#
+
+-- source include/master-slave.inc
+-- source include/have_debug.inc
+
+# make the test only run once (STMT is actually needed because we rely
+# on SHOW PROCESS LIST output in some of the tests)
+-- source include/have_binlog_format_statement.inc
+
+# todo: sleep() is deterministic!!!
+call mtr.add_suppression('Unsafe statement written to the binary log using statement format since BINLOG_FORMAT = STATEMENT');
+
+# restart slave in MTS mode
+connection slave;
+source include/stop_slave.inc;
+set @save.mts_slave_parallel_workers= @@global.mts_slave_parallel_workers;
+set @@global.mts_slave_parallel_workers= 4;
+set @save.mts_checkpoint_period= @@global.mts_checkpoint_period;
+set @@global.mts_checkpoint_period= 500; # 0.5 sec
+# to avoid warnings
+set @save.slave_transaction_retries= @@global.slave_transaction_retries;
+
+#
+# Idea of demonstration.
+#
+# Two transactions on the same db are generated for MTS. The first has some
+# natural delay, and the 2nd will be locked out of its data.
+# While the first is being processed Coordinator empties the relay-log
+# to enter waiting for more events.
+# It does so being awakened @@global.slave_checkpoint_period-ically
+# to check out if some groups of events have been done to update SBM (demo 1).
+#
+# Resetting of SBM is a possibility whenever Coordinator
+# notices no more groups left neither to read nor to process (demo 2).
+#
+
+start slave io_thread;
+source include/wait_for_slave_io_to_start.inc;
+
+connection master;
+
+create table t1 (f1 int) engine=innodb;
+create table t2 (f1 int) engine=innodb;
+save_master_pos;
+
+# 1st transaction that will be used to compute SBM.
+# It will hang $idle time at least in processing to get min estimate for SBM
+let $idle= 3;
+
+begin;
+ eval insert into t1 values (sleep($idle) + 1);
+commit;
+
+# 2nd transaction to block Coordinator from resetting SBM
+begin;
+ insert into t2 values (1);
+commit;
+
+connection slave;
+source include/start_slave.inc;
+
+#
+# all events are in relay-log
+#
+connection master;
+
+source include/sync_slave_io_with_master.inc;
+
+# tables are created on slave
+connection slave;
+sync_with_master;
+
+connection slave1;
+lock table t2 write; # to block 2nd trans
+
+connection slave;
+
+let $count= 1;
+let $table= t1;
+source include/wait_until_rows_count.inc;
+
+--echo First transaction is done, now get Seconds_Behind_Master after it...
+
+#
+# demo 1:
+# So now 1st is over, and it SBM must have gain at least $idle seconds.
+#
+let $slave_param= Seconds_Behind_Master;
+let $slave_param_comparison= >=;
+let $slave_param_value= $idle;
+source include/wait_for_slave_param.inc;
+
+# checking sanity of SBM value
+let $sbm= query_get_value("SHOW SLAVE STATUS", Seconds_Behind_Master, 1);
+let $lower_bound= $idle;
+if (`SELECT $sbm < $idle`)
+{
+ -- echo *** Seconds behind master is $sbm less than minimum expected $idle ***
+ -- source include/show_rpl_debug_info.inc
+ -- die
+}
+
+--echo Seconds_Behind_Master after first transaction is as expected.
+
+connection slave1;
+unlock tables;
+
+#
+# demo 2: Resetting of SBM upon all groups have been processed and
+# nothing left in relay-log.
+#
+connection slave;
+let $slave_param= Seconds_Behind_Master;
+let $slave_param_comparison= =;
+let $slave_param_value= 0;
+source include/wait_for_slave_param.inc;
+
+##
+# cleanup
+##
+connection master;
+drop tables t1, t2;
+
+sync_slave_with_master;
+
+set @@global.mts_checkpoint_period= @save.mts_checkpoint_period;
+set @@global.mts_slave_parallel_workers= @save.mts_slave_parallel_workers;
+set @@global.slave_transaction_retries= @save.slave_transaction_retries;
+
+--source include/rpl_end.inc
+
+--echo end of the tests
=== modified file 'mysql-test/suite/rpl/t/rpl_seconds_behind_master.test'
--- a/mysql-test/suite/rpl/t/rpl_seconds_behind_master.test 2010-12-19 17:22:30 +0000
+++ b/mysql-test/suite/rpl/t/rpl_seconds_behind_master.test 2011-06-10 08:04:00 +0000
@@ -5,6 +5,10 @@
# on SHOW PROCESS LIST output in some of the tests)
-- source include/have_binlog_format_statement.inc
+# this test logics requires adaption to MTS policies for SBM
+# (see rpl_parallel_seconds_behind_master.test)
+-- source include/not_mts_slave_parallel_workers.inc
+
SET @old_debug= @@global.debug;
call mtr.add_suppression('Unsafe statement binlogged in statement format since BINLOG_FORMAT = STATEMENT');
=== modified file 'sql/log_event.cc'
--- a/sql/log_event.cc 2011-06-08 20:18:08 +0000
+++ b/sql/log_event.cc 2011-06-10 08:04:00 +0000
@@ -2635,6 +2635,7 @@ Slave_worker *Log_event::get_slave_worke
ret_worker->checkpoint_notified= TRUE;
}
ptr_g->checkpoint_seqno= rli->checkpoint_seqno;
+ ptr_g->ts= when + (time_t) exec_time;
rli->checkpoint_seqno++;
}
@@ -6514,17 +6515,17 @@ int Rotate_log_event::do_update_pos(Rela
((!rli->is_parallel_exec() && !rli->is_in_group()) ||
rli->mts_group_status != Relay_log_info::MTS_IN_GROUP))
{
+ if (rli->is_parallel_exec())
+ {
+ (void) mts_checkpoint_routine(rli, 0, FALSE, TRUE); // todo: error branch
+ }
+
mysql_mutex_lock(&rli->data_lock);
DBUG_PRINT("info", ("old group_master_log_name: '%s' "
"old group_master_log_pos: %lu",
rli->get_group_master_log_name(),
(ulong) rli->get_group_master_log_pos()));
- if (rli->is_parallel_exec())
- {
- (void) mts_checkpoint_routine(rli, 0, FALSE, TRUE); // todo: error branch
- }
-
memcpy((void *)rli->get_group_master_log_name(),
new_log_ident, ident_len + 1);
rli->notify_group_master_log_name_update();
@@ -6537,7 +6538,7 @@ int Rotate_log_event::do_update_pos(Rela
mysql_mutex_unlock(&rli->data_lock);
rli->flush_info(TRUE); // todo: error branch
if (rli->is_parallel_exec())
- rli->reset_notified_checkpoint(0);
+ rli->reset_notified_checkpoint(0, when + (time_t) exec_time);
/*
Reset thd->variables.option_bits and sql_mode etc, because this could be the signal of
=== modified file 'sql/rpl_rli.cc'
--- a/sql/rpl_rli.cc 2011-06-08 20:18:08 +0000
+++ b/sql/rpl_rli.cc 2011-06-10 08:04:00 +0000
@@ -191,8 +191,14 @@ void Relay_log_info::reset_notified_rela
Worker notices the new checkpoint value at the group commit
to reset the current bitmap and starts using the clean bitmap
indexed from zero of being reset checkpoint_seqno.
+ New seconds_behind_master timestamp is installed.
+
+ @param shift number of bits to shift by Worker due to the current
+ checkpoint change.
+ @param new_ts new seconds_behind_master timestamp value unless zero.
+ Zero could be due to FD event.
*/
-void Relay_log_info::reset_notified_checkpoint(ulong shift)
+void Relay_log_info::reset_notified_checkpoint(ulong shift, time_t new_ts= 0)
{
if (!is_parallel_exec())
return;
@@ -207,6 +213,13 @@ void Relay_log_info::reset_notified_chec
w->bitmap_shifted += shift; // to reset at passing the accumulate value into GAQ
}
checkpoint_seqno= 0;
+
+ if (new_ts)
+ {
+ mysql_mutex_lock(&data_lock);
+ last_master_timestamp= new_ts;
+ mysql_mutex_unlock(&data_lock);
+ }
}
static inline int add_relay_log(Relay_log_info* rli,LOG_INFO* linfo)
=== modified file 'sql/rpl_rli.h'
--- a/sql/rpl_rli.h 2011-06-08 20:18:08 +0000
+++ b/sql/rpl_rli.h 2011-06-10 08:04:00 +0000
@@ -537,7 +537,7 @@ public:
Coordinator notifies Workers about this event. Coordinator and Workers
maintain a bitmap of executed group that is reset with a new checkpoint.
*/
- void reset_notified_checkpoint(ulong);
+ void reset_notified_checkpoint(ulong, time_t);
/**
Helper function to do after statement completion.
=== modified file 'sql/rpl_rli_pdb.h'
--- a/sql/rpl_rli_pdb.h 2011-06-08 20:18:08 +0000
+++ b/sql/rpl_rli_pdb.h 2011-06-10 08:04:00 +0000
@@ -144,7 +144,8 @@ typedef struct st_slave_job_group
my_off_t checkpoint_relay_log_pos; // T-event lop_pos filled by W for CheckPoint
char* checkpoint_relay_log_name;
volatile uchar done; // Flag raised by W, read and reset by C
- ulong shifted; // shift the last CP bitmap at receiving a new CP
+ ulong shifted; // shift the last CP bitmap at receiving a new CP
+ time_t ts; // Group's timestampt to update Seconds_behind_master
} Slave_job_group;
#define retrieve_job(from, to) \
=== modified file 'sql/rpl_slave.cc'
--- a/sql/rpl_slave.cc 2011-06-08 20:18:08 +0000
+++ b/sql/rpl_slave.cc 2011-06-10 08:04:00 +0000
@@ -3088,7 +3088,8 @@ static int exec_relay_log_event(THD* thd
event) or ev->when is set to 0, or a FD from master, we don't update the
last_master_timestamp.
*/
- if (!(ev->is_artificial_event() || ev->is_relay_log_event() ||
+ if (!(rli->is_parallel_exec() ||
+ ev->is_artificial_event() || ev->is_relay_log_event() ||
(ev->when == 0) || ev->get_type_code() == FORMAT_DESCRIPTION_EVENT))
{
rli->last_master_timestamp= ev->when + (time_t) ev->exec_time;
@@ -4153,7 +4154,7 @@ bool mts_checkpoint_routine(Relay_log_in
} // end of commit_positions
*/
- rli->reset_notified_checkpoint(cnt);
+ rli->reset_notified_checkpoint(cnt, rli->gaq->lwm.ts);
end:
set_timespec_nsec(rli->last_clock, 0);
@@ -6005,7 +6006,8 @@ static Log_event* next_event(Relay_log_i
/* shows zero while it is sleeping (and until the next event
is about to be executed) */
- rli->last_master_timestamp= 0;
+ if (!rli->is_parallel_exec())
+ rli->last_master_timestamp= 0;
DBUG_ASSERT(rli->relay_log.get_open_count() ==
rli->cur_log_old_open_count);
@@ -6089,6 +6091,8 @@ static Log_event* next_event(Relay_log_i
mysql_mutex_unlock(log_lock);
mts_checkpoint_routine(rli, period, FALSE, FALSE); // TODO: ALFRANIO ERROR
mysql_mutex_lock(log_lock);
+ if (rli->gaq->empty())
+ rli->last_master_timestamp= 0;
set_timespec_nsec(waittime, period);
thd->enter_cond(log_cond, log_lock,
Attachment: [text/bzr-bundle] bzr/andrei.elkin@oracle.com-20110610080400-yo3gawj9c42sq8wm.bundle
| Thread |
|---|
| • bzr commit into mysql-next-mr-wl5569 branch (andrei.elkin:3284) WL#5569 | Andrei Elkin | 10 Jun |
