| Server IP : 172.67.216.182 / Your IP : 172.69.176.77 Web Server : Apache System : Linux krdc-ubuntu-s-2vcpu-4gb-amd-blr1-01.localdomain 5.15.0-142-generic #152-Ubuntu SMP Mon May 19 10:54:31 UTC 2025 x86_64 User : www ( 1000) PHP Version : 7.4.33 Disable Function : passthru,exec,system,putenv,chroot,chgrp,chown,shell_exec,popen,proc_open,pcntl_exec,ini_alter,ini_restore,dl,openlog,syslog,readlink,symlink,popepassthru,pcntl_alarm,pcntl_fork,pcntl_waitpid,pcntl_wait,pcntl_wifexited,pcntl_wifstopped,pcntl_wifsignaled,pcntl_wifcontinued,pcntl_wexitstatus,pcntl_wtermsig,pcntl_wstopsig,pcntl_signal,pcntl_signal_dispatch,pcntl_get_last_error,pcntl_strerror,pcntl_sigprocmask,pcntl_sigwaitinfo,pcntl_sigtimedwait,pcntl_exec,pcntl_getpriority,pcntl_setpriority,imap_open,apache_setenv MySQL : OFF | cURL : ON | WGET : ON | Perl : ON | Python : OFF | Sudo : ON | Pkexec : ON Directory : /www/server/mysql/src/sql/ |
Upload File : |
/* Copyright (c) 2010, 2023, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License, version 2.0,
as published by the Free Software Foundation.
This program is also distributed with certain software (including
but not limited to OpenSSL) that is licensed under separate terms,
as designated in a particular file or component or in included license
documentation. The authors of MySQL hereby grant you an additional
permission to link the program and your derivative works with the
separately licensed software that they have included with MySQL.
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, version 2.0, 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
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */
#include "rpl_info_factory.h"
#include "log.h" // sql_print_error
#include "mysqld.h" // key_master_info_run_lock
#include "rpl_info_dummy.h" // Rpl_info_dummy
#include "rpl_info_file.h" // Rpl_info_file
#include "rpl_info_table.h" // Rpl_info_table
#include "rpl_info_table_access.h" // Rpl_info_table_access
#include "rpl_mi.h" // Master_info
#include "rpl_msr.h" // channel_map
#include "rpl_rli.h" // Relay_log_info
#include "rpl_rli_pdb.h" // Slave_worker
/*
Defines meta information on diferent repositories.
*/
Rpl_info_factory::struct_table_data Rpl_info_factory::rli_table_data;
Rpl_info_factory::struct_file_data Rpl_info_factory::rli_file_data;
Rpl_info_factory::struct_table_data Rpl_info_factory::mi_table_data;
Rpl_info_factory::struct_file_data Rpl_info_factory::mi_file_data;
Rpl_info_factory::struct_file_data Rpl_info_factory::worker_file_data;
Rpl_info_factory::struct_table_data Rpl_info_factory::worker_table_data;
/**
Creates a Master info repository whose type is defined as a parameter.
@param[in] mi_option Type of the repository, e.g. FILE TABLE.
@param[in] channel the channel for which mi is to be created
@param[in] to_decide_repo the flag is set to true if mi repositories
are allowed to convert. For details,
see init_slave()
The execution fails if a user requests a type but a different type
already exists in the system. This is done to avoid that a user
accidentally accesses the wrong repository and makes the slave go out
of sync.
@retval Pointer to Master_info Success
@retval NULL Failure
*/
Master_info *Rpl_info_factory::create_mi(uint mi_option, const char* channel,
bool to_decide_repo)
{
Master_info* mi= NULL;
Rpl_info_handler* handler_src= NULL;
Rpl_info_handler* handler_dest= NULL;
uint instances= 1;
const char *msg= "Failed to allocate memory for the master info "
"structure";
DBUG_ENTER("Rpl_info_factory::create_mi");
if (!(mi= new Master_info(
#ifdef HAVE_PSI_INTERFACE
&key_master_info_run_lock,
&key_master_info_data_lock,
&key_master_info_sleep_lock,
&key_master_info_thd_lock,
&key_master_info_data_cond,
&key_master_info_start_cond,
&key_master_info_stop_cond,
&key_master_info_sleep_cond,
#endif
instances, channel
)))
goto err;
if(init_repositories(mi_table_data, mi_file_data, mi_option, instances,
&handler_src, &handler_dest, &msg))
goto err;
if(to_decide_repo)
{
if (decide_repository(mi, mi_option, &handler_src, &handler_dest, &msg))
goto err;
}
else
{
if (handler_dest->get_rpl_info_type() != INFO_REPOSITORY_TABLE)
{
sql_print_error("Slave: Wrong repository. Respository should be TABLE");
goto err;
}
mi->set_rpl_info_handler(handler_dest);
if (mi->set_info_search_keys(handler_dest))
goto err;
delete handler_src;
}
DBUG_RETURN(mi);
err:
delete handler_src;
delete handler_dest;
if (mi)
{
/*
The handler was previously deleted so we need to remove
any reference to it.
*/
mi->set_rpl_info_handler(NULL);
mi->channel_wrlock();
delete mi;
}
sql_print_error("Error creating master info: %s.", msg);
DBUG_RETURN(NULL);
}
/**
Allows to change the master info repository after startup.
@param[in] mi Reference to Master_info.
@param[in] mi_option Type of the repository, e.g. FILE TABLE.
@param[out] msg Error message if something goes wrong.
@retval FALSE No error
@retval TRUE Failure
*/
bool Rpl_info_factory::change_mi_repository(Master_info *mi,
uint mi_option,
const char **msg)
{
Rpl_info_handler* handler_src= mi->get_rpl_info_handler();
Rpl_info_handler* handler_dest= NULL;
uint instances= 1;
DBUG_ENTER("Rpl_info_factory::change_mi_repository");
assert(handler_src);
if (init_repositories(mi_table_data, mi_file_data, mi_option, instances,
NULL, &handler_dest, msg))
goto err;
if (decide_repository(mi, mi_option, &handler_src, &handler_dest, msg))
goto err;
DBUG_RETURN(FALSE);
err:
delete handler_dest;
handler_dest= NULL;
sql_print_error("Error changing the type of master info's repository: %s.", *msg);
DBUG_RETURN(TRUE);
}
/**
Creates a Relay log info repository whose type is defined as a parameter.
@param[in] rli_option Type of the Relay log info repository
@param[in] is_slave_recovery If the slave should try to start a recovery
process to get consistent relay log files
@param[in] channel the channel for which mi is to be created
@param[in] to_decide_repo If true, rli repositories are allowed
to convert from one repo to other
The execution fails if a user requests a type but a different type
already exists in the system. This is done to avoid that a user
accidentally accesses the wrong repository and make the slave go out
of sync.
@retval Pointer to Relay_log_info Success
@retval NULL Failure
*/
Relay_log_info *Rpl_info_factory::create_rli(uint rli_option,
bool is_slave_recovery,
const char* channel,
bool to_decide_repo)
{
Relay_log_info *rli= NULL;
Rpl_info_handler* handler_src= NULL;
Rpl_info_handler* handler_dest= NULL;
uint instances= 1;
uint worker_repository= INVALID_INFO_REPOSITORY;
uint worker_instances= 1;
const char *msg= NULL;
const char *msg_alloc= "Failed to allocate memory for the relay log info "
"structure";
DBUG_ENTER("Rpl_info_factory::create_rli");
/*
Returns how many occurrences of worker's repositories exist. For example,
if the repository is a table, this retrieves the number of rows in it.
Besides, it also returns the type of the repository where entries were
found.
*/
if (rli_option != INFO_REPOSITORY_DUMMY &&
scan_repositories(&worker_instances, &worker_repository,
worker_table_data, worker_file_data, &msg))
goto err;
if (!(rli= new Relay_log_info(is_slave_recovery
#ifdef HAVE_PSI_INTERFACE
,&key_relay_log_info_run_lock,
&key_relay_log_info_data_lock,
&key_relay_log_info_sleep_lock,
&key_relay_log_info_thd_lock,
&key_relay_log_info_data_cond,
&key_relay_log_info_start_cond,
&key_relay_log_info_stop_cond,
&key_relay_log_info_sleep_cond
#endif
, instances, channel,
(rli_option != INFO_REPOSITORY_TABLE
&& rli_option != INFO_REPOSITORY_FILE)
)))
{
msg= msg_alloc;
goto err;
}
if(init_repositories(rli_table_data, rli_file_data, rli_option, instances,
&handler_src, &handler_dest, &msg))
goto err;
if (rli_option != INFO_REPOSITORY_DUMMY &&
worker_repository != INVALID_INFO_REPOSITORY &&
worker_repository != rli_option)
{
opt_rli_repository_id= rli_option= worker_repository;
sql_print_warning("It is not possible to change the type of the relay log "
"repository because there are workers repositories with "
"possible execution gaps. "
"The value of --relay_log_info_repository is altered to "
"one of the found Worker repositories. "
"The gaps have to be sorted out before resuming with "
"the type change.");
std::swap(handler_src, handler_dest);
}
if (to_decide_repo)
{
if (decide_repository(rli, rli_option, &handler_src, &handler_dest, &msg))
goto err;
}
else
{
if(channel != NULL)
{
/* Here dest code should be TABLE type repo. See, init_slave() */
if (handler_dest->get_rpl_info_type() != INFO_REPOSITORY_TABLE)
{
sql_print_error("Slave: Wrong repository. Repository should be TABLE");
goto err;
}
if(rli->set_info_search_keys(handler_dest))
goto err;
}
rli->set_rpl_info_handler(handler_dest);
/*
By this time, rli must be loaded with it's primary key,
which is channel_name
*/
delete handler_src;
}
DBUG_RETURN(rli);
err:
delete handler_src;
delete handler_dest;
if (rli)
{
/*
The handler was previously deleted so we need to remove
any reference to it.
*/
rli->set_rpl_info_handler(NULL);
delete rli;
}
sql_print_error("Error creating relay log info: %s.", msg);
DBUG_RETURN(NULL);
}
/**
Allows to change the relay log info repository after startup.
@param[in] mi Pointer to Relay_log_info.
@param[in] mi_option Type of the repository, e.g. FILE TABLE.
@param[out] msg Error message if something goes wrong.
@retval FALSE No error
@retval TRUE Failure
*/
bool Rpl_info_factory::change_rli_repository(Relay_log_info *rli,
uint rli_option,
const char **msg)
{
Rpl_info_handler* handler_src= rli->get_rpl_info_handler();
Rpl_info_handler* handler_dest= NULL;
uint instances= 1;
DBUG_ENTER("Rpl_info_factory::change_rli_repository");
assert(handler_src != NULL);
if (init_repositories(rli_table_data, rli_file_data, rli_option,
instances, NULL, &handler_dest, msg))
goto err;
if (decide_repository(rli, rli_option, &handler_src, &handler_dest,
msg))
goto err;
DBUG_RETURN(FALSE);
err:
delete handler_dest;
handler_dest= NULL;
sql_print_error("Error changing the type of relay log info's repository: %s.", *msg);
DBUG_RETURN(TRUE);
}
/**
Delete all info from Worker info tables to render them useless in
future MTS recovery, and indicate that in Coordinator info table.
@return false on success, true when a failure in deletion or writing
to Coordinator table fails.
*/
bool Rpl_info_factory::reset_workers(Relay_log_info *rli)
{
bool error= true;
DBUG_ENTER("Rpl_info_factory::reset_workers");
if (rli->recovery_parallel_workers == 0)
DBUG_RETURN(0);
if (Rpl_info_file::do_reset_info(Slave_worker::get_number_worker_fields(),
worker_file_data.pattern,
worker_file_data.name_indexed))
goto err;
if (Rpl_info_table::do_reset_info(Slave_worker::get_number_worker_fields(),
MYSQL_SCHEMA_NAME.str, WORKER_INFO_NAME.str,
rli->channel,
Slave_worker::get_channel_field_index()))
goto err;
error= false;
DBUG_EXECUTE_IF("mts_debug_reset_workers_fails", error= true;);
err:
if (error)
sql_print_error("Could not delete from Slave Workers info repository.");
rli->recovery_parallel_workers= 0;
rli->clear_mts_recovery_groups();
if (rli->flush_info(true))
{
error= true;
sql_print_error("Could not store the reset Slave Worker state into "
"the slave info repository.");
}
DBUG_RETURN(error);
}
/**
Creates a Slave worker repository whose type is defined as a parameter.
@param[in] rli_option Type of the repository, e.g. FILE TABLE.
@param[in] rli Pointer to Relay_log_info.
The execution fails if a user requests a type but a different type
already exists in the system. This is done to avoid that a user
accidentally accesses the wrong repository and make the slave go out
of sync.
@retval Pointer to Slave_worker Success
@retval NULL Failure
*/
Slave_worker *Rpl_info_factory::create_worker(uint rli_option, uint worker_id,
Relay_log_info *rli,
bool is_gaps_collecting_phase)
{
Rpl_info_handler* handler_src= NULL;
Rpl_info_handler* handler_dest= NULL;
Slave_worker* worker= NULL;
const char *msg= "Failed to allocate memory for the worker info "
"structure";
DBUG_ENTER("Rpl_info_factory::create_worker");
/*
Define the name of the worker and its repository.
*/
char *pos= my_stpcpy(worker_file_data.name, worker_file_data.pattern);
sprintf(pos, "%u", worker_id + 1);
if (!(worker= new Slave_worker(rli
#ifdef HAVE_PSI_INTERFACE
,&key_relay_log_info_run_lock,
&key_relay_log_info_data_lock,
&key_relay_log_info_sleep_lock,
&key_relay_log_info_thd_lock,
&key_relay_log_info_data_cond,
&key_relay_log_info_start_cond,
&key_relay_log_info_stop_cond,
&key_relay_log_info_sleep_cond
#endif
, worker_id, rli->get_channel()
)))
goto err;
if(init_repositories(worker_table_data, worker_file_data, rli_option,
worker_id + 1, &handler_src, &handler_dest, &msg))
goto err;
/*
Preparing the being set up handler with search keys early.
The file repo type handler can't be manupulated this way and it does
not have to.
*/
if (handler_dest->get_rpl_info_type() == INFO_REPOSITORY_TABLE)
worker->set_info_search_keys(handler_dest);
/* get_num_instances() requires channel_map lock */
/*
assert(channel_map.get_num_instances() <= 1 ||
(rli_option == 1 && handler_dest->get_rpl_info_type() == 1));
*/
if (decide_repository(worker, rli_option, &handler_src, &handler_dest, &msg))
goto err;
if (DBUG_EVALUATE_IF("mts_worker_thread_init_fails", 1, 0) ||
worker->rli_init_info(is_gaps_collecting_phase))
{
DBUG_EXECUTE_IF("enable_mts_worker_failure_init",
{
DBUG_SET("-d,mts_worker_thread_init_fails");
DBUG_SET("-d,enable_mts_worker_failure_init");
});
DBUG_EXECUTE_IF("enable_mts_wokrer_failure_in_recovery_finalize",
{
DBUG_SET("-d,mts_worker_thread_init_fails");
DBUG_SET("-d,enable_mts_wokrer_failure_in_recovery_finalize");
});
msg= "Failed to initialize the worker info structure";
goto err;
}
if (rli->info_thd && rli->info_thd->is_error())
{
msg= "Failed to initialize worker info table";
goto err;
}
DBUG_RETURN(worker);
err:
delete handler_src;
delete handler_dest;
if (worker)
{
/*
The handler was previously deleted so we need to remove
any reference to it.
*/
worker->set_rpl_info_handler(NULL);
delete worker;
}
sql_print_error("Error creating relay log info: %s.", msg);
DBUG_RETURN(NULL);
}
static void build_worker_info_name(char* to,
const char* path,
const char* fname)
{
assert(to);
char* pos= to;
if (path[0])
pos= my_stpcpy(pos, path);
pos= my_stpcpy(pos, "worker-");
pos= my_stpcpy(pos, fname);
my_stpcpy(pos, ".");
}
/**
Initializes startup information on diferent repositories.
*/
void Rpl_info_factory::init_repository_metadata()
{
/* Needed for the file names and paths for worker info files. */
size_t len;
char* relay_log_info_file_name;
char relay_log_info_file_dirpart[FN_REFLEN];
/* Extract the directory name from relay_log_info_file */
dirname_part(relay_log_info_file_dirpart, relay_log_info_file, &len);
relay_log_info_file_name= relay_log_info_file + len;
rli_table_data.n_fields= Relay_log_info::get_number_info_rli_fields();
rli_table_data.schema= MYSQL_SCHEMA_NAME.str;
rli_table_data.name= RLI_INFO_NAME.str;
rli_table_data.n_pk_fields= 0;
rli_table_data.pk_field_indexes= NULL;
rli_file_data.n_fields= Relay_log_info::get_number_info_rli_fields();
my_stpcpy(rli_file_data.name, relay_log_info_file);
my_stpcpy(rli_file_data.pattern, relay_log_info_file);
rli_file_data.name_indexed= false;
mi_table_data.n_fields= Master_info::get_number_info_mi_fields();
mi_table_data.schema= MYSQL_SCHEMA_NAME.str;
mi_table_data.name= MI_INFO_NAME.str;
mi_table_data.n_pk_fields= 1;
mi_table_data.pk_field_indexes= Master_info::get_table_pk_field_indexes();
mi_file_data.n_fields= Master_info::get_number_info_mi_fields();
my_stpcpy(mi_file_data.name, master_info_file);
my_stpcpy(mi_file_data.pattern, master_info_file);
rli_file_data.name_indexed= false;
worker_table_data.n_fields= Slave_worker::get_number_worker_fields();
worker_table_data.schema= MYSQL_SCHEMA_NAME.str;
worker_table_data.name= WORKER_INFO_NAME.str;
worker_table_data.n_pk_fields= 2;
worker_table_data.pk_field_indexes=
Slave_worker::get_table_pk_field_indexes();
worker_file_data.n_fields= Slave_worker::get_number_worker_fields();
build_worker_info_name(worker_file_data.name,
relay_log_info_file_dirpart,
relay_log_info_file_name);
build_worker_info_name(worker_file_data.pattern,
relay_log_info_file_dirpart,
relay_log_info_file_name);
worker_file_data.name_indexed= true;
}
/**
Decides during startup what repository will be used based on the following
decision table:
\code
|--------------+-----------------------+-----------------------|
| Exists \ Opt | SOURCE | DESTINATION |
|--------------+-----------------------+-----------------------|
| ~is_s, ~is_d | - | Create/Update D |
| ~is_s, is_d | - | Continue with D |
| is_s, ~is_d | Copy S into D | Create/Update D |
| is_s, is_d | Error | Error |
|--------------+-----------------------+-----------------------|
\endcode
@param[in] info Either master info or relay log info.
@param[in] option Identifies the type of the repository that will
be used, i.e., destination repository.
@param[out] handler_src Source repository from where information is
copied into the destination repository.
@param[out] handler_dest Destination repository to where informaiton is
copied.
@param[out] msg Error message if something goes wrong.
@retval FALSE No error
@retval TRUE Failure
*/
bool Rpl_info_factory::decide_repository(Rpl_info *info, uint option,
Rpl_info_handler **handler_src,
Rpl_info_handler **handler_dest,
const char **msg)
{
bool error= true;
enum_return_check return_check_src= ERROR_CHECKING_REPOSITORY;
enum_return_check return_check_dst= ERROR_CHECKING_REPOSITORY;
DBUG_ENTER("Rpl_info_factory::decide_repository");
if (option == INFO_REPOSITORY_DUMMY)
{
delete (*handler_src);
*handler_src= NULL;
info->set_rpl_info_handler(*handler_dest);
error = false;
goto err;
}
assert((*handler_src) != NULL && (*handler_dest) != NULL &&
(*handler_src) != (*handler_dest));
return_check_src= check_src_repository(info, option, handler_src);
return_check_dst= (*handler_dest)->do_check_info(info->get_internal_id()); // approx via scan, not field_values! Todo: reconsider any need for that at least in the Worker case
if (return_check_src == ERROR_CHECKING_REPOSITORY ||
return_check_dst == ERROR_CHECKING_REPOSITORY)
{
/*
If there is a problem with one of the repositories we print out
more information and exit.
*/
DBUG_RETURN(check_error_repository(info, *handler_src, *handler_dest,
return_check_src,
return_check_dst, msg));
}
else
{
if ((return_check_src == REPOSITORY_EXISTS &&
return_check_dst == REPOSITORY_DOES_NOT_EXIST) ||
(return_check_src == REPOSITORY_EXISTS &&
return_check_dst == REPOSITORY_EXISTS))
{
/*
If there is no error, we can proceed with the normal operation.
However, if both repositories are set an error will be printed
out.
*/
if (return_check_src == REPOSITORY_EXISTS &&
return_check_dst == REPOSITORY_EXISTS)
{
*msg= "Multiple replication metadata repository instances "
"found with data in them. Unable to decide which is "
"the correct one to choose";
goto err;
}
/*
Do a low-level initialization to be able to do a state transfer.
*/
if (init_repositories(info, handler_src, handler_dest, msg))
goto err;
/*
Transfer information from source to destination and delete the
source. Note this is not fault-tolerant and a crash before removing
source may cause the next restart to fail as is_src and is_dest may
be true. Moreover, any failure in removing the source may lead to
the same.
/Alfranio
*/
if (info->copy_info(*handler_src, *handler_dest) ||
(*handler_dest)->flush_info(true))
{
*msg= "Error transfering information";
goto err;
}
(*handler_src)->end_info();
if ((*handler_src)->remove_info())
{
*msg= "Error removing old repository";
goto err;
}
}
else if (return_check_src == REPOSITORY_DOES_NOT_EXIST &&
return_check_dst == REPOSITORY_EXISTS)
{
assert(info->get_rpl_info_handler() == NULL);
if ((*handler_dest)->do_init_info(info->get_internal_id()))
{
*msg= "Error reading repository";
goto err;
}
}
else
{
assert(return_check_src == REPOSITORY_DOES_NOT_EXIST &&
return_check_dst == REPOSITORY_DOES_NOT_EXIST);
info->inited= false;
}
delete (*handler_src);
*handler_src= NULL;
info->set_rpl_info_handler(*handler_dest);
error= false;
}
err:
DBUG_RETURN(error);
}
/**
This method is called by the decide_repository() and is used to check if
the source repository exits.
@param[in] info Either master info or relay log info.
@param[in] option Identifies the type of the repository that will
be used, i.e., destination repository.
@param[out] handler_src Source repository from where information is
@return enum_return_check The repository's status.
*/
enum_return_check
Rpl_info_factory::check_src_repository(Rpl_info *info,
uint option,
Rpl_info_handler **handler_src)
{
enum_return_check return_check_src= ERROR_CHECKING_REPOSITORY;
bool live_migration = info->get_rpl_info_handler() != NULL;
if (!live_migration)
{
/*
This is not a live migration and we don't know whether the repository
exists or not.
*/
return_check_src= (*handler_src)->do_check_info(info->get_internal_id());
/*
Since this is not a live migration, if we are using file repository
and there is some error on table repository (for instance, engine
disabled) we can ignore it instead of stopping replication.
A warning saying that table is not ready to be used was logged.
*/
if (ERROR_CHECKING_REPOSITORY == return_check_src &&
INFO_REPOSITORY_FILE == option &&
INFO_REPOSITORY_TABLE == (*handler_src)->do_get_rpl_info_type())
{
return_check_src= REPOSITORY_DOES_NOT_EXIST;
/*
If a already existent thread was used to access info tables,
current_thd will point to it and we must clear access error on
it. If a temporary thread was used, then there is nothing to
clean because the thread was already deleted.
See Rpl_info_table_access::create_thd().
*/
if (current_thd)
current_thd->clear_error();
}
}
else
{
/*
This is a live migration as the repository is already associated to.
However, we cannot assume that it really exists, for instance, if a
file was really created.
This situation may happen when we start a slave for the first time
but skips its initialization and tries to migrate it.
*/
return_check_src= (*handler_src)->do_check_info();
}
return return_check_src;
}
/**
This method is called by the decide_repository() and is used print out
information on errors.
@param info Either master info or relay log info.
@param handler_src Source repository from where information is
copied into the destination repository.
@param handler_dest Destination repository to where informaiton is
copied.
@param err_src Possible error status of the source repo check
@param err_dst Possible error status of the destination repo check
@param[out] msg Error message if something goes wrong.
@retval TRUE Failure
*/
bool Rpl_info_factory::check_error_repository(Rpl_info *info,
Rpl_info_handler *handler_src,
Rpl_info_handler *handler_dest,
enum_return_check err_src,
enum_return_check err_dst,
const char **msg)
{
bool error = true;
/*
If there is an error in any of the source or destination
repository checks, the normal operation can't be proceeded.
The runtime repository won't be initialized.
*/
if (err_src == ERROR_CHECKING_REPOSITORY)
sql_print_error("Error in checking %s repository info type of %s.",
handler_src->get_description_info(),
handler_src->get_rpl_info_type_str());
if (err_dst == ERROR_CHECKING_REPOSITORY)
sql_print_error("Error in checking %s repository info type of %s.",
handler_dest->get_description_info(),
handler_dest->get_rpl_info_type_str());
*msg= "Error checking repositories";
return error;
}
/**
This method is called by the decide_repository() and is used to initialize
the repositories through a low-level interfacei, which means that if they
do not exist nothing will be created.
@param[in] info Either master info or relay log info.
@param[out] handler_src Source repository from where information is
copied into the destination repository.
@param[out] handler_dest Destination repository to where informaiton is
copied.
@param[out] msg Error message if something goes wrong.
@retval FALSE No error
@retval TRUE Failure
*/
bool Rpl_info_factory::init_repositories(Rpl_info *info,
Rpl_info_handler **handler_src,
Rpl_info_handler **handler_dest,
const char **msg)
{
bool live_migration = info->get_rpl_info_handler() != NULL;
if (!live_migration)
{
if ((*handler_src)->do_init_info(info->get_internal_id()) ||
(*handler_dest)->do_init_info(info->get_internal_id()))
{
*msg= "Error transfering information";
return true;
}
}
else
{
if ((*handler_dest)->do_init_info(info->get_internal_id()))
{
*msg= "Error transfering information";
return true;
}
}
return false;
}
/**
Creates repositories that will be associated to either the Master_info
or Relay_log_info.
@param[in] table_data Defines information to create a table repository.
@param[in] file_data Defines information to create a file repository.
@param[in] rep_option Identifies the type of the repository that will
be used, i.e., destination repository.
@param[in] instance Identifies the instance of the repository that
will be used.
@param[out] handler_src Source repository from where information is
copied into the destination repository.
@param[out] handler_dest Destination repository to where informaiton is
copied.
@param[out] msg Error message if something goes wrong.
@retval FALSE No error
@retval TRUE Failure
*/
bool Rpl_info_factory::init_repositories(const struct_table_data table_data,
const struct_file_data file_data,
uint rep_option,
uint instance,
Rpl_info_handler **handler_src,
Rpl_info_handler **handler_dest,
const char **msg)
{
bool error= TRUE;
*msg= "Failed to allocate memory for master info repositories";
DBUG_ENTER("Rpl_info_factory::init_mi_repositories");
assert(handler_dest != NULL);
switch (rep_option)
{
case INFO_REPOSITORY_FILE:
if (!(*handler_dest= new Rpl_info_file(file_data.n_fields,
file_data.pattern,
file_data.name,
file_data.name_indexed)))
goto err;
if (handler_src &&
!(*handler_src= new Rpl_info_table(table_data.n_fields,
table_data.schema,
table_data.name,
table_data.n_pk_fields,
table_data.pk_field_indexes)))
goto err;
break;
case INFO_REPOSITORY_TABLE:
if (!(*handler_dest= new Rpl_info_table(table_data.n_fields,
table_data.schema,
table_data.name,
table_data.n_pk_fields,
table_data.pk_field_indexes)))
goto err;
if (handler_src &&
!(*handler_src= new Rpl_info_file(file_data.n_fields,
file_data.pattern,
file_data.name,
file_data.name_indexed)))
goto err;
break;
case INFO_REPOSITORY_DUMMY:
if (!(*handler_dest= new Rpl_info_dummy(Master_info::get_number_info_mi_fields())))
goto err;
break;
default:
assert(0);
}
error= FALSE;
err:
DBUG_RETURN(error);
}
bool Rpl_info_factory::scan_repositories(uint* found_instances,
uint* found_rep_option,
const struct_table_data table_data,
const struct_file_data file_data,
const char **msg)
{
bool error= false;
uint file_instances= 0;
uint table_instances= 0;
assert(found_rep_option != NULL);
DBUG_ENTER("Rpl_info_factory::scan_repositories");
if (Rpl_info_table::do_count_info(table_data.n_fields, table_data.schema,
table_data.name, &table_instances))
{
error= true;
goto err;
}
if (Rpl_info_file::do_count_info(file_data.n_fields, file_data.pattern,
file_data.name_indexed, &file_instances))
{
error= true;
goto err;
}
if (file_instances != 0 && table_instances != 0)
{
error= true;
*msg= "Multiple repository instances found with data in "
"them. Unable to decide which is the correct one to "
"choose";
goto err;
}
if (table_instances != 0)
{
*found_instances= table_instances;
*found_rep_option= INFO_REPOSITORY_TABLE;
}
else if (file_instances != 0)
{
*found_instances= file_instances;
*found_rep_option= INFO_REPOSITORY_FILE;
}
else
{
*found_instances= 0;
*found_rep_option= INVALID_INFO_REPOSITORY;
}
err:
DBUG_RETURN(error);
}
/**
This function should be called from init_slave() only.
During the server start, read all the slave repositories
on disk (either in FILE or TABLE form) and create corresponding
slave info objects. Each thus created master_info object is
added to pchannel_map.
Multisource replication is supported by only TABLE based
repositories. Based on this fact, the following table shows
the supported cases considering the repository type and
multiple channels of a slave.
Each <---> represents a channel with a name on top of it.
'' is an empty stringed channel (or default channel).
'N' indicates some name for a channel.
+-----------------------------+------------------+-----------+
| channels | Supported? FILE | TABLE |
+-----------------------------+------------------+-----------+
| '' | | |
| A) Master<------->Slave | YES | YES |
| | | |
| | | |
| 'N' | | |
| B) Master<------->Slave | NO | YES |
| | | |
| '' | | |
| Master0<------------+ | | |
| 'N' v | NO | YES |
| C) Master1<----------->Slave| | |
| 'N' ^ | | |
| Mastern<------------+ | | |
| | | |
| | | |
| 'N' | | |
| Master1<------------+ | | |
| 'N' v | NO | YES |
| D) Master2<----------->Slave| | |
| 'N' ^ | | |
| Mastern<------------+ | | |
| | | |
| | | |
+-----------------------------+------------------+-----------+
In a new server, A) shown above is created by default.
If there are multiple 'named' channels, but and if a default_channel
is not created, it is created.
Some points to note from the above table
=========================================
From the table it also follows that conversion of repositories
is possible *ONLY* in the case of A) i.e for ex: if B) type repository
(i.e a named slave channel) was found during server starup but the user
repository option is INFO_REPOSITORY_FILE, then we exit the function.
@note: only for type A) i.e default channel, it is permissable to
have different repo types for Master_info and Relay_log_info
(Ex: FILE for mi and TABLE for rli)
@note: The above restrictions break factory pattern in the code
which has been followed throughout before.
@note: All the repository conversion(or live migration) functions
(ex: decide_repository()) take Rpl_info::internal_id as an
identifier which is always 1 for the case of Master_info and
Relay_log_info. So, in the case of multisource replication,
the decision to convert the repositories shall be made even before
invoking decide_repository(). In other words, if a channel is not a
default channel('') we shall not invoke decide_repository().
@note: In general, the algorithm in creation of slave info object is:
l1: new slave_info;
l2: Initialize the repository handlers
l3: if (default_channel)
check and convert repositories
else
// TABLE type repository
set the value of PK in the TABLE handler.
@note: Update from 5.6 to 5.7(which has Channel_Name in slave_info_tables)
is handled in the upgrade script as usual.
@param[in] mi_option the user provided master_info_repository
@param[in] rli_option the user provided relay_log_info_repository
@param[in] thread_mask thread mask
@param[in] pchannel_map the pointer to the multi source map
(see, rpl_msr.h)
@return
@retval false success
@retval true fail
*/
bool Rpl_info_factory::create_slave_info_objects(uint mi_option,
uint rli_option,
int thread_mask,
Multisource_info *pchannel_map)
{
DBUG_ENTER("create_slave_info_objects");
Master_info* mi= NULL;
const char* msg= NULL;
bool error= false, channel_error;
bool default_channel_existed_previously= false;
std::vector<std::string> channel_list;
/* Number of instances of Master_info repository */
uint mi_instances= 0;
/* At this point, the repository in invalid or unknown */
uint mi_repository= INVALID_INFO_REPOSITORY;
/*
Number of instances of Relay_log_info_repository.
(Number of Slave worker objects that will be created by the Coordinator
(when slave_parallel_workers>0) at a later stage and not here).
*/
uint rli_instances= 0;
/* At this point, the repository is invalid or unknown */
uint rli_repository= INVALID_INFO_REPOSITORY;
/*
Initialize the repository metadata. This metadata is the
name of files to look in case of FILE type repository, and the
names of table to look in case of TABLE type repository.
*/
Rpl_info_factory::init_repository_metadata();
/* Count the number of Master_info and Relay_log_info repositories */
if (scan_repositories(&mi_instances, &mi_repository, mi_table_data,
mi_file_data, &msg) ||
scan_repositories(&rli_instances, &rli_repository, rli_table_data,
rli_file_data, &msg))
{
/* msg will contain the reason of failure */
sql_print_error("Slave: %s", msg);
error= true;
goto end;
}
/* Make a list of all channels if the slave was connected to previously*/
if (load_channel_names_from_repository(channel_list,
mi_instances,
mi_repository,
pchannel_map->get_default_channel(),
&default_channel_existed_previously))
{
sql_print_error("Slave: Could not create channel list");
error= true;
goto end;
}
if ((mi_option == INFO_REPOSITORY_FILE ||
rli_option == INFO_REPOSITORY_FILE) && channel_list.size() > 1)
{
/* Not supported cases of B) C) and D) above */
sql_print_error("Slave: This slave was a multisourced slave previously which"
" is supported only by both TABLE based master info and relay"
" log info repositories. Found one or both of the info repos"
" to be type FILE. Set both repos to type TABLE.");
error= true;
goto end;
}
/* Adding the default channel if needed. */
if (!default_channel_existed_previously)
{
std::string str(pchannel_map->get_default_channel());
channel_list.push_back(str);
}
/*
Create and initialize the channels.
Even if there is an error during one channel creation, we continue to
iterate until we have created the other channels.
For compatibility reasons, we have to separate the print out
of the error messages.
*/
for (std::vector<std::string>::iterator it= channel_list.begin();
it != channel_list.end(); ++it)
{
const char* cname= (*it).c_str();
bool is_default_channel= !strcmp(cname, pchannel_map->get_default_channel());
channel_error= !(mi= create_mi_and_rli_objects(mi_option, rli_option, cname,
(channel_list.size() == 1) ? 1 : 0,
pchannel_map));
/*
Read the channel configuration from the repository if the channel name
was read from the repository.
*/
if (!channel_error && (!is_default_channel || default_channel_existed_previously))
{
bool ignore_if_no_info= (channel_list.size() == 1) ? true : false;
channel_error= load_mi_and_rli_from_repositories(mi,
ignore_if_no_info,
thread_mask);
}
if (channel_error)
{
sql_print_error("Slave: Failed to initialize the master info structure"
" for channel '%s'; its record may still be present in"
" 'mysql.slave_master_info' table, consider "
"deleting it.", cname);
}
error= error || channel_error;
}
end:
DBUG_RETURN(error);
}
/**
Create Master_info and Relay_log_info objects for a new channel.
Also, set cross dependencies between these objects used all over
the code.
Both master_info and relay_log_info repositories should be of the type
TABLE. We do a check for this here as well.
@param[in] mi_option master info repository
@param[in] rli_option relay log info repository
@param[in] channel the channel for which these objects
should be created.
@param[in] to_decide_repo For this channel, check if repositories
are allowed to convert from one type to other.
@param[in] pchannel_map a pointer to channel_map
@return Pointer pointer to the created Master_info
@return NULL when creation fails
*/
Master_info*
Rpl_info_factory::create_mi_and_rli_objects(uint mi_option,
uint rli_option,
const char* channel,
bool to_decide_repo,
Multisource_info* pchannel_map)
{
DBUG_ENTER("Rpl_info_factory::create_mi_and_rli_objects");
Master_info *mi= NULL;
Relay_log_info *rli= NULL;
if (!(mi= Rpl_info_factory::create_mi(mi_option, channel, to_decide_repo)))
DBUG_RETURN(NULL);
if (!(rli= Rpl_info_factory::create_rli(rli_option, relay_log_recovery,
channel, to_decide_repo)))
{
mi->channel_wrlock();
delete mi;
mi= NULL;
DBUG_RETURN(NULL);
}
/* Set the cross dependencies used all over the code */
mi->set_relay_log_info(rli);
rli->set_master_info(mi);
/* Add to multisource map*/
if (pchannel_map->add_mi(channel, mi))
{
mi->channel_wrlock();
delete mi;
delete rli;
DBUG_RETURN(NULL);
}
DBUG_RETURN(mi);
}
/**
Make a list of all the channels if existed on the previos slave run.
@param[out] channel_list the names of all channels that exists
on this slave.
@param[in] mi_instances number of master_info repositories
@param[in] mi_repository Found master_info repository
@param[in] default_channel pointer to default channel.
@param[out] default_channel_existed_previously
Value filled with true if default channel
existed previously. False if it is not.
@return
@retval true fail
@retval false success
*/
bool
Rpl_info_factory::load_channel_names_from_repository(std::vector<std::string>& channel_list,
uint mi_instances,
uint mi_repository,
const char* default_channel,
bool *default_channel_existed_previously)
{
DBUG_ENTER("Rpl_info_factory::load_channel_names_from_repository");
*default_channel_existed_previously= false;
switch(mi_repository)
{
case INFO_REPOSITORY_FILE:
assert(mi_instances == 1);
/* insert default channel */
{
std::string str(default_channel);
channel_list.push_back(str);
}
*default_channel_existed_previously= true;
break;
case INFO_REPOSITORY_TABLE:
if (load_channel_names_from_table(channel_list, default_channel,
default_channel_existed_previously))
DBUG_RETURN(true);
break;
case INVALID_INFO_REPOSITORY:
/* file and table instanaces are zero, nothing to be done*/
break;
default:
assert(0);
}
DBUG_RETURN(false);
}
/**
In a multisourced slave, during init_slave(), the repositories
are read to initialize the slave info objects. To initialize
the slave info objects, we need the number of channels the slave
was connected to previously. The following function, finds the
number of channels in the master info repository.
Later, this chanenl list is used to create a pair of {mi, rli}
objects required for IO and SQL threads respectively.
@param [out] channel_list A reference to the channel list.
This will be filled after reading the
master info table, row by row.
@param[in] default_channel pointer to default channel.
@param[out] default_channel_existed_previously
Value filled with true if default channel
existed previously. False if it is not.
@todo: Move it to Rpl_info_table and make it generic to extract
all the PK list from the tables (but it not yet necessary)
*/
bool
Rpl_info_factory::load_channel_names_from_table(std::vector<std::string> &channel_list,
const char* default_channel,
bool *default_channel_existed_previously)
{
DBUG_ENTER(" Rpl_info_table::load_channel_names_from_table");
int error= 1;
TABLE *table= 0;
ulong saved_mode;
Open_tables_backup backup;
Rpl_info_table* info= 0;
THD *thd= 0;
char buff[MAX_FIELD_WIDTH];
*default_channel_existed_previously= false;
String str(buff, sizeof(buff), system_charset_info); // to extract channel names
uint channel_field= Master_info::get_channel_field_num() -1;
if(!(info= new Rpl_info_table(mi_table_data.n_fields,
mi_table_data.schema, mi_table_data.name,
mi_table_data.n_pk_fields,
mi_table_data.pk_field_indexes)))
DBUG_RETURN(true);
thd= info->access->create_thd();
saved_mode= thd->variables.sql_mode;
// Ensure channel names are always read consistently
thd->variables.sql_mode &= ~MODE_PAD_CHAR_TO_FULL_LENGTH;
/*
Opens and locks the rpl_info table before accessing it.
*/
if (info->access->open_table(thd, info->str_schema, info->str_table,
info->get_number_info(), TL_READ, &table,
&backup))
{
/*
We cannot simply print out a warning message at this
point because this may represent a bootstrap.
*/
error= 0;
goto err;
}
/* Do ha_handler random init for full scanning */
if ((error= table->file->ha_rnd_init(true)))
DBUG_RETURN(true);
/* Ensure that the table pk (Channel_name) is at the correct position */
if (info->verify_table_primary_key_fields(table))
{
sql_print_error("Slave: Failed to create a channel from master info "
"table repository.");
error= -1;
goto err;
}
/*
Load all the values in record[0] for each row
and then extract channel name from it
*/
do
{
error= table->file->ha_rnd_next(table->record[0]);
switch(error)
{
case 0:
/* extract the channel name from table->field and append to the list */
table->field[channel_field]->val_str(&str);
channel_list.push_back(std::string(str.c_ptr_safe()));
if (!strcmp(str.c_ptr_safe(), default_channel))
*default_channel_existed_previously= true;
break;
case HA_ERR_END_OF_FILE:
break;
default:
DBUG_PRINT("info", ("Failed to get next record"
" (ha_rnd_next returns %d)", error));
}
}
while(!error);
/*close the table */
err:
table->file->ha_rnd_end();
info->access->close_table(thd, table, &backup, error);
thd->variables.sql_mode= saved_mode;
info->access->drop_thd(thd);
delete info;
DBUG_RETURN(error != HA_ERR_END_OF_FILE && error != 0);
}