#At file:///home/acorreia/workspace.oracle/repository.mysql/bzrwork/wl-5569/mysql-next-mr-wl5569/ based on revid:andrei.elkin@stripped
3315 Alfranio Correia 2011-06-22
Fixed how mts copes with recovery.
modified:
sql/log_event.cc
sql/rpl_rli.cc
sql/rpl_rli.h
sql/rpl_slave.cc
=== modified file 'sql/log_event.cc'
--- a/sql/log_event.cc 2011-06-21 14:05:32 +0000
+++ b/sql/log_event.cc 2011-06-22 17:54:23 +0000
@@ -2864,19 +2864,14 @@ int Log_event::apply_event(Relay_log_inf
{
bool skip= bitmap_is_set(&c_rli->recovery_groups, c_rli->mts_recovery_index);
- if (ends_group()) // TODO: || ! seen_begin
- {
- c_rli->mts_recovery_index++;
- if (--c_rli->mts_recovery_group_cnt == 0)
- {
- c_rli->recovery_parallel_workers= c_rli->slave_parallel_workers;
- c_rli->mts_recovery_index= 0;
- }
- }
if (skip)
+ {
DBUG_RETURN(0);
- else
+ }
+ else
+ {
DBUG_RETURN(do_apply_event(rli));
+ }
}
if (!(parallel= rli->is_parallel_exec()) ||
=== modified file 'sql/rpl_rli.cc'
--- a/sql/rpl_rli.cc 2011-06-18 18:58:21 +0000
+++ b/sql/rpl_rli.cc 2011-06-22 17:54:23 +0000
@@ -75,7 +75,8 @@ Relay_log_info::Relay_log_info(bool is_s
this_worker(NULL), slave_parallel_workers(0),
recovery_parallel_workers(0),
checkpoint_group(mts_checkpoint_group), mts_recovery_group_cnt(0),
- mts_recovery_index(0), mts_group_status(MTS_NOT_IN_GROUP),
+ mts_recovery_index(0), mts_recovery_group_seen_begin(0),
+ mts_group_status(MTS_NOT_IN_GROUP),
sql_delay(0), sql_delay_end(0), m_flags(0)
{
DBUG_ENTER("Relay_log_info::Relay_log_info");
=== modified file 'sql/rpl_rli.h'
--- a/sql/rpl_rli.h 2011-06-14 18:23:13 +0000
+++ b/sql/rpl_rli.h 2011-06-22 17:54:23 +0000
@@ -472,6 +472,7 @@ public:
MY_BITMAP recovery_groups; // bitmap used during recovery
ulong mts_recovery_group_cnt; // number of groups to execute at recovery
ulong mts_recovery_index; // running index of recoverable groups
+ bool mts_recovery_group_seen_begin;
/*
temporary tables are held by Coordinator though are created and dropped
explicilty by Workers. The following lock has to be taken by either party
=== modified file 'sql/rpl_slave.cc'
--- a/sql/rpl_slave.cc 2011-06-21 14:05:32 +0000
+++ b/sql/rpl_slave.cc 2011-06-22 17:54:23 +0000
@@ -2724,6 +2724,7 @@ int apply_event_and_update_pos(Log_event
int exec_res= 0;
bool skip_event= FALSE;
Log_event *ev= *ptr_ev;
+ Log_event::enum_skip_reason reason= Log_event::EVENT_SKIP_NOT;
DBUG_ENTER("apply_event_and_update_pos");
@@ -2766,7 +2767,18 @@ int apply_event_and_update_pos(Log_event
ev->when= my_time(0);
ev->thd = thd; // because up to this point, ev->thd == 0
- int reason= ev->shall_skip(rli);
+ if (!(rli->is_mts_recovery() && bitmap_is_set(&rli->recovery_groups,
+ rli->mts_recovery_index)))
+ {
+ reason= ev->shall_skip(rli);
+ }
+
+ DBUG_PRINT("Mts", ("Mts is recovering %d, number of bits set %d, "
+ "bitmap is set %d, index %lu.",
+ rli->is_mts_recovery(), bitmap_bits_set(&rli->recovery_groups),
+ bitmap_is_set(&rli->recovery_groups, rli->mts_recovery_index),
+ rli->mts_recovery_index));
+
if (reason == Log_event::EVENT_SKIP_COUNT)
{
sql_slave_skip_counter= --rli->slave_skip_counter;
@@ -2777,6 +2789,7 @@ int apply_event_and_update_pos(Log_event
// Sleeps if needed, and unlocks rli->data_lock.
if (sql_delay_event(ev, thd, rli))
DBUG_RETURN(0);
+
exec_res= ev->apply_event(rli);
if (!exec_res && (ev->worker != rli))
@@ -2858,7 +2871,25 @@ int apply_event_and_update_pos(Log_event
}
else
mysql_mutex_unlock(&rli->data_lock);
-
+
+ if (rli->is_mts_recovery())
+ {
+ if (ev->starts_group())
+ {
+ rli->mts_recovery_group_seen_begin= TRUE;
+ }
+ if (ev->ends_group() || !rli->mts_recovery_group_seen_begin)
+ {
+ rli->mts_recovery_index++;
+ if (--rli->mts_recovery_group_cnt == 0)
+ {
+ rli->recovery_parallel_workers= rli->slave_parallel_workers;
+ rli->mts_recovery_index= 0;
+ }
+ rli->mts_recovery_group_seen_begin= FALSE;
+ }
+ }
+
DBUG_PRINT("info", ("apply_event error = %d", exec_res));
if (exec_res == 0)
{
@@ -3982,14 +4013,16 @@ bool mts_recovery_groups(Relay_log_info
}
if (ev->ends_group() || !curr_group_seen_begin)
{
- int ret;
+ int ret= 0;
LOG_POS_COORD ev_coord= { (char *) rli->get_group_master_log_name(),
ev->log_pos };
+ curr_group_seen_begin= FALSE;
+ rli->mts_recovery_group_cnt++;
+
sql_print_information("Group Recoverying relay log info "
"group_master_log_name %s, event_master_log_pos %lu.",
rli->group_master_log_name,
(ulong) ev->log_pos);
-
if ((ret= mts_event_coord_cmp(&ev_coord, &w_last)) == 0)
{
/*
@@ -4007,6 +4040,10 @@ bool mts_recovery_groups(Relay_log_info
ALFRANIO --> Add more debug messages.
*/
DBUG_PRINT("Mts-checkpoint", ("Bit %u is set.", i));
+ }
+ else
+ {
+ DBUG_PRINT("Mts-checkpoint", ("Bit %u is not set.", i));
}
}
#endif
@@ -4014,11 +4051,12 @@ bool mts_recovery_groups(Relay_log_info
ALFRANIO --> Add more debug messages.
*/
DBUG_PRINT("Mts-checkpoint",
- ("Ignoring the first #(%lu) bits. Doing a shift.",
- rli->mts_recovery_group_cnt));
+ ("Doing a shift ini(%lu) end(%lu).",
+ (w->checkpoint_seqno + 1) - rli->mts_recovery_group_cnt,
+ w->checkpoint_seqno));
- for (uint i= (w->checkpoint_seqno - rli->mts_recovery_group_cnt) + 1, j= 0;
- i <= w->checkpoint_seqno; i++, j++)
+ for (uint i= (w->checkpoint_seqno + 1) - rli->mts_recovery_group_cnt,
+ j= 0; i <= w->checkpoint_seqno; i++, j++)
{
if (bitmap_is_set(&w->group_execed, i))
{
@@ -4029,7 +4067,6 @@ bool mts_recovery_groups(Relay_log_info
bitmap_fast_test_and_set(groups, j);
}
}
- rli->mts_recovery_group_cnt++;
delete ev;
ev= NULL;
break;
@@ -4037,7 +4074,6 @@ bool mts_recovery_groups(Relay_log_info
else
{
DBUG_ASSERT(ret < 0);
- rli->mts_recovery_group_cnt++;
}
}
delete ev;
Attachment: [text/bzr-bundle] bzr/alfranio.correia@oracle.com-20110622175423-myzbr6j5koflw2k7.bundle
| Thread |
|---|
| • bzr commit into mysql-next-mr-wl5569 branch (alfranio.correia:3315) | Alfranio Correia | 23 Jun |
