reckless slave option

bison 1.34 yacc rule compliance
more robust handling of relay log corruption

mysql-test/r/rpl000014.result

@@ -8,7 +8,7 @@ File	Position	Binlog_do_db	Binlog_ignore_db
 master-bin.001	79		
 show slave status;
 Master_Host	Master_User	Master_Port	Connect_retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_do_db	Replicate_ignore_db	Last_errno	Last_error	Skip_counter	Exec_master_log_pos	Relay_log_space
-127.0.0.1	root	MASTER_PORT	1	master-bin.001	79	slave-relay-bin.002	120	master-bin.001	Yes	Yes			0		0	79	124
+127.0.0.1	root	MASTER_PORT	1	master-bin.001	79	slave-relay-bin.002	120	master-bin.001	Yes	Yes			0		0	79	128
 change master to master_log_pos=73;
 slave stop;
 change master to master_log_pos=73;

mysql-test/r/rpl_log.result

@@ -75,7 +75,7 @@ slave-bin.002	115	Query	1	62	use test; insert into t1 values (1)
 slave-bin.002	175	Query	1	122	use test; drop table t1
 show slave status;
 Master_Host	Master_User	Master_Port	Connect_retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_do_db	Replicate_ignore_db	Last_errno	Last_error	Skip_counter	Exec_master_log_pos	Relay_log_space
-127.0.0.1	root	MASTER_PORT	1	master-bin.002	170	slave-relay-bin.002	916	master-bin.002	Yes	Yes			0		0	170	920
+127.0.0.1	root	MASTER_PORT	1	master-bin.002	170	slave-relay-bin.002	916	master-bin.002	Yes	Yes			0		0	170	924
 show new master for slave with master_log_file='master-bin.001' and
 master_log_pos=4 and  master_server_id=1;
 Log_name	Log_pos

sql/mysqld.cc

@@ -227,6 +227,11 @@ SHOW_COMP_OPTION have_query_cache=SHOW_OPTION_NO;
 #endif
 
 bool opt_skip_slave_start = 0; // If set, slave is not autostarted
+
+/* if set, some standard measures to enforce
+   slave data intergity will not be performed
+ */
+bool opt_reckless_slave = 0; 
 static bool opt_do_pstack = 0;
 static ulong opt_specialflag=SPECIAL_ENGLISH;
 static ulong back_log,connect_timeout,concurrency;
@@ -2740,6 +2745,7 @@ enum options {
 	       OPT_RPL_RECOVERY_RANK,OPT_INIT_RPL_ROLE,
 	       OPT_RELAY_LOG, OPT_RELAY_LOG_INDEX, OPT_RELAY_LOG_INFO_FILE,
                OPT_SLAVE_SKIP_ERRORS, OPT_DES_KEY_FILE, OPT_LOCAL_INFILE,
+	       OPT_RECKLESS_SLAVE,
 	       OPT_SSL_SSL, OPT_SSL_KEY, OPT_SSL_CERT, OPT_SSL_CA,
 	       OPT_SSL_CAPATH, OPT_SSL_CIPHER
 };
@@ -2850,6 +2856,7 @@ static struct option long_options[] = {
 #endif
   {"pid-file",              required_argument, 0, (int) OPT_PID_FILE},
   {"port",                  required_argument, 0, 'P'},
+  {"reckless-slave",        no_argument,       0, (int) OPT_RECKLESS_SLAVE},
   {"replicate-do-db",       required_argument, 0, (int) OPT_REPLICATE_DO_DB},
   {"replicate-do-table",       required_argument, 0,
    (int) OPT_REPLICATE_DO_TABLE},
@@ -3938,6 +3945,10 @@ static void get_options(int argc,char **argv)
       opt_slow_log=1;
       opt_slow_logname=optarg;
       break;
+    case (int)OPT_RECKLESS_SLAVE:
+      opt_reckless_slave = 1;
+      init_slave_skip_errors("all");
+      break;
     case (int)OPT_SKIP_SLAVE_START:
       opt_skip_slave_start = 1;
       break;

sql/slave.cc

@@ -191,11 +191,22 @@ int init_relay_log_pos(RELAY_LOG_INFO* rli,const char* log,
     pos = rli->relay_log_pos; // already inited
   else
     rli->relay_log_pos = pos;
-  if (rli->relay_log.find_first_log(&rli->linfo,log))
+
+  // test to see if the previous run was with the skip of purging
+  // if yes, we do not purge when we restart
+  if (rli->relay_log.find_first_log(&rli->linfo,""))
   {
     *errmsg="Could not find first log during relay log initialization";
     goto err;
   }
+  if (strcmp(log,rli->linfo.log_file_name))
+    rli->skip_log_purge=1;
+  
+  if (rli->relay_log.find_first_log(&rli->linfo,log))
+  {
+    *errmsg="Could not find target log during relay log initialization";
+    goto err;
+  }
   strnmov(rli->relay_log_name,rli->linfo.log_file_name,
 	  sizeof(rli->relay_log_name));
   // to make end_io_cache(&rli->cache_buf) safe in all cases
@@ -2497,6 +2508,15 @@ Log_event* next_event(RELAY_LOG_INFO* rli)
       return ev;
     }
     DBUG_ASSERT(thd==rli->sql_thd);
+    if (opt_reckless_slave)
+      cur_log->error = 0;
+    if ( cur_log->error < 0)
+    {
+      errmsg = "slave SQL thread aborted because of I/O error";
+      goto err;
+    }
+    
+
     if (!cur_log->error) /* EOF */
     {
       /*
@@ -2605,12 +2625,12 @@ event(errno=%d,cur_log->error=%d)",
 		      my_errno,cur_log->error);
       // set read position to the beginning of the event
       my_b_seek(cur_log,rli->relay_log_pos+rli->pending);
-      // no need to hog the mutex while we sleep
-      pthread_mutex_unlock(&rli->data_lock);
-      safe_sleep(rli->sql_thd,1,(CHECK_KILLED_FUNC)sql_slave_killed,
-		 (void*)rli);
-      pthread_mutex_lock(&rli->data_lock);
+      /* otherwise, we have had a partial read */
+      /* TODO; see if there is a way to do this without this goto */
+      errmsg = "Aborting slave SQL thread because of partial event read";
+      goto err;
     }
+
   }
   if (!errmsg && was_killed)
     errmsg = "slave SQL thread was killed";

sql/slave.h

@@ -30,7 +30,7 @@ extern bool use_slave_mask;
 extern char* slave_load_tmpdir;
 extern my_string master_info_file,relay_log_info_file;
 extern my_string opt_relay_logname, opt_relaylog_index_name;
-extern bool opt_skip_slave_start;
+extern bool opt_skip_slave_start, opt_reckless_slave;
 extern ulong relay_log_space_limit;
 struct st_master_info;