1 package DBIx::Class::Storage::DBI::Replicated;
7 require DBIx::Class::Optional::Dependencies;
8 if ( my $missing = DBIx::Class::Optional::Dependencies->req_missing_for('replicated') ) {
9 die "The following modules are required for Replicated storage support: $missing\n";
14 use DBIx::Class::Storage::DBI;
15 use DBIx::Class::Storage::DBI::Replicated::Pool;
16 use DBIx::Class::Storage::DBI::Replicated::Balancer;
17 use DBIx::Class::Storage::DBI::Replicated::Types qw/BalancerClassNamePart DBICSchema DBICStorageDBI/;
18 use MooseX::Types::Moose qw/ClassName HashRef Object/;
19 use Scalar::Util 'reftype';
21 use List::Util qw/min max reduce/;
22 use Context::Preserve 'preserve_context';
25 use namespace::clean -except => 'meta';
29 DBIx::Class::Storage::DBI::Replicated - BETA Replicated database support
33 The Following example shows how to change an existing $schema to a replicated
34 storage type, add some replicated (read-only) databases, and perform reporting
37 You should set the 'storage_type attribute to a replicated type. You should
38 also define your arguments, such as which balancer you want and any arguments
39 that the Pool object should get.
41 my $schema = Schema::Class->clone;
42 $schema->storage_type(['::DBI::Replicated', { balancer_type => '::Random' }]);
43 $schema->connection(...);
45 Next, you need to add in the Replicants. Basically this is an array of
46 arrayrefs, where each arrayref is database connect information. Think of these
47 arguments as what you'd pass to the 'normal' $schema->connect method.
49 $schema->storage->connect_replicants(
50 [$dsn1, $user, $pass, \%opts],
51 [$dsn2, $user, $pass, \%opts],
52 [$dsn3, $user, $pass, \%opts],
55 Now, just use the $schema as you normally would. Automatically all reads will
56 be delegated to the replicants, while writes to the master.
58 $schema->resultset('Source')->search({name=>'etc'});
60 You can force a given query to use a particular storage using the search
61 attribute 'force_pool'. For example:
63 my $rs = $schema->resultset('Source')->search(undef, {force_pool=>'master'});
65 Now $rs will force everything (both reads and writes) to use whatever was setup
66 as the master storage. 'master' is hardcoded to always point to the Master,
67 but you can also use any Replicant name. Please see:
68 L<DBIx::Class::Storage::DBI::Replicated::Pool> and the replicants attribute for more.
70 Also see transactions and L</execute_reliably> for alternative ways to
71 force read traffic to the master. In general, you should wrap your statements
72 in a transaction when you are reading and writing to the same tables at the
73 same time, since your replicants will often lag a bit behind the master.
75 If you have a multi-statement read only transaction you can force it to select
76 a random server in the pool by:
78 my $rs = $schema->resultset('Source')->search( undef,
79 { force_pool => $db->storage->read_handler->next_storage }
84 Warning: This class is marked BETA. This has been running a production
85 website using MySQL native replication as its backend and we have some decent
86 test coverage but the code hasn't yet been stressed by a variety of databases.
87 Individual DBs may have quirks we are not aware of. Please use this in first
88 development and pass along your experiences/bug fixes.
90 This class implements replicated data store for DBI. Currently you can define
91 one master and numerous slave database connections. All write-type queries
92 (INSERT, UPDATE, DELETE and even LAST_INSERT_ID) are routed to master
93 database, all read-type queries (SELECTs) go to the slave database.
95 Basically, any method request that L<DBIx::Class::Storage::DBI> would normally
96 handle gets delegated to one of the two attributes: L</read_handler> or to
97 L</write_handler>. Additionally, some methods need to be distributed
98 to all existing storages. This way our storage class is a drop in replacement
99 for L<DBIx::Class::Storage::DBI>.
101 Read traffic is spread across the replicants (slaves) occurring to a user
102 selected algorithm. The default algorithm is random weighted.
106 The consistency between master and replicants is database specific. The Pool
107 gives you a method to validate its replicants, removing and replacing them
108 when they fail/pass predefined criteria. Please make careful use of the ways
109 to force a query to run against Master when needed.
113 Replicated Storage has additional requirements not currently part of
114 L<DBIx::Class>. See L<DBIx::Class::Optional::Dependencies> for more details.
118 This class defines the following attributes.
122 The underlying L<DBIx::Class::Schema> object this storage is attaching
135 Contains the classname which will instantiate the L</pool> object. Defaults
136 to: L<DBIx::Class::Storage::DBI::Replicated::Pool>.
143 default=>'DBIx::Class::Storage::DBI::Replicated::Pool',
145 'create_pool' => 'new',
151 Contains a hashref of initialized information to pass to the Balancer object.
152 See L<DBIx::Class::Storage::DBI::Replicated::Pool> for available arguments.
166 The replication pool requires a balance class to provider the methods for
167 choose how to spread the query load across each replicant in the pool.
171 has 'balancer_type' => (
173 isa=>BalancerClassNamePart,
176 default=> 'DBIx::Class::Storage::DBI::Replicated::Balancer::First',
178 'create_balancer' => 'new',
184 Contains a hashref of initialized information to pass to the Balancer object.
185 See L<DBIx::Class::Storage::DBI::Replicated::Balancer> for available arguments.
189 has 'balancer_args' => (
199 Is a L<DBIx::Class::Storage::DBI::Replicated::Pool> or derived class. This is a
200 container class for one or more replicated databases.
206 isa=>'DBIx::Class::Storage::DBI::Replicated::Pool',
217 Is a L<DBIx::Class::Storage::DBI::Replicated::Balancer> or derived class. This
218 is a class that takes a pool (L<DBIx::Class::Storage::DBI::Replicated::Pool>)
224 isa=>'DBIx::Class::Storage::DBI::Replicated::Balancer',
226 handles=>[qw/auto_validate_every/],
231 The master defines the canonical state for a pool of connected databases. All
232 the replicants are expected to match this databases state. Thus, in a classic
233 Master / Slaves distributed system, all the slaves are expected to replicate
234 the Master's state as quick as possible. This is the only database in the
235 pool of databases that is allowed to handle write traffic.
245 =head1 ATTRIBUTES IMPLEMENTING THE DBIx::Storage::DBI INTERFACE
247 The following methods are delegated all the methods required for the
248 L<DBIx::Class::Storage::DBI> interface.
252 my $method_dispatch = {
264 deployment_statements
267 build_datetime_parser
282 with_deferred_fk_checks
290 relname_to_table_alias
292 _default_dbi_connect_attributes
294 _dbic_connect_attributes
300 bind_attribute_by_data_type
304 _dbh_execute_for_fetch
306 _dbh_execute_inserts_with_no_binds
307 _select_args_to_query
310 _normalize_connect_info
324 /, Class::MOP::Class->initialize('DBIx::Class::Storage::DBIHacks')->get_method_list ],
329 _dbh_columns_info_for
332 unimplemented => [qw/
333 _arm_global_destructor
336 get_use_dbms_capability
337 set_use_dbms_capability
343 _determine_connector_driver
344 _extract_driver_from_connect_info
346 _warn_undetermined_driver
353 _perform_autoinc_retrieval
354 _autoinc_supplied_for_op
367 # the capability framework
368 # not sure if CMOP->initialize does evil things to DBIC::S::DBI, fix if a problem
370 { $_ =~ /^ _ (?: use | supports | determine_supports ) _ /x }
371 ( Class::MOP::Class->initialize('DBIx::Class::Storage::DBI')->get_all_method_names )
375 if (DBIx::Class::_ENV_::DBICTEST) {
378 for my $type (keys %$method_dispatch) {
379 for (@{$method_dispatch->{$type}}) {
380 push @{$seen->{$_}}, $type;
384 if (my @dupes = grep { @{$seen->{$_}} > 1 } keys %$seen) {
386 'The following methods show up multiple times in ::Storage::DBI::Replicated handlers:',
387 (map { "$_: " . (join ', ', @{$seen->{$_}}) } sort @dupes),
392 if (my @cant = grep { ! DBIx::Class::Storage::DBI->can($_) } keys %$seen) {
394 '::Storage::DBI::Replicated specifies handling of the following *NON EXISTING* ::Storage::DBI methods:',
401 for my $method (@{$method_dispatch->{unimplemented}}) {
402 __PACKAGE__->meta->add_method($method, sub {
404 $self->throw_exception("$method() must not be called on ".(blessed $self).' objects');
410 Defines an object that implements the read side of L<DBIx::Class::Storage::DBI>.
414 has 'read_handler' => (
418 handles=>$method_dispatch->{reader},
423 Defines an object that implements the write side of L<DBIx::Class::Storage::DBI>,
424 as well as methods that don't write or read that can be called on only one
425 storage, methods that return a C<$dbh>, and any methods that don't make sense to
430 has 'write_handler' => (
434 handles=>$method_dispatch->{writer},
439 has _master_connect_info_opts =>
440 (is => 'rw', isa => HashRef, default => sub { {} });
442 =head2 around: connect_info
444 Preserves master's C<connect_info> options (for merging with replicants.)
445 Also sets any Replicated-related options from connect_info, such as
446 C<pool_type>, C<pool_args>, C<balancer_type> and C<balancer_args>.
450 around connect_info => sub {
451 my ($next, $self, $info, @extra) = @_;
453 $self->throw_exception(
454 'connect_info can not be retrieved from a replicated storage - '
455 . 'accessor must be called on a specific pool instance'
456 ) unless defined $info;
458 my $merge = Hash::Merge->new('LEFT_PRECEDENT');
461 for my $arg (@$info) {
462 next unless (reftype($arg)||'') eq 'HASH';
463 %opts = %{ $merge->merge($arg, \%opts) };
467 if (@opts{qw/pool_type pool_args/}) {
468 $self->pool_type(delete $opts{pool_type})
472 $merge->merge((delete $opts{pool_args} || {}), $self->pool_args)
475 ## Since we possibly changed the pool_args, we need to clear the current
476 ## pool object so that next time it is used it will be rebuilt.
480 if (@opts{qw/balancer_type balancer_args/}) {
481 $self->balancer_type(delete $opts{balancer_type})
482 if $opts{balancer_type};
484 $self->balancer_args(
485 $merge->merge((delete $opts{balancer_args} || {}), $self->balancer_args)
488 $self->balancer($self->_build_balancer)
492 $self->_master_connect_info_opts(\%opts);
494 return preserve_context {
495 $self->$next($info, @extra);
497 # Make sure master is blessed into the correct class and apply role to it.
498 my $master = $self->master;
499 $master->_determine_driver;
500 Moose::Meta::Class->initialize(ref $master);
502 DBIx::Class::Storage::DBI::Replicated::WithDSN->meta->apply($master);
504 # link pool back to master
505 $self->pool->master($master);
511 This class defines the following methods.
515 L<DBIx::Class::Schema> when instantiating its storage passed itself as the
516 first argument. So we need to massage the arguments a bit so that all the
517 bits get put into the correct places.
522 my ($class, $schema, $storage_type_args, @args) = @_;
533 Lazy builder for the L</master> attribute.
539 my $master = DBIx::Class::Storage::DBI->new($self->schema);
545 Lazy builder for the L</pool> attribute.
551 $self->create_pool(%{$self->pool_args});
554 =head2 _build_balancer
556 Lazy builder for the L</balancer> attribute. This takes a Pool object so that
557 the balancer knows which pool it's balancing.
561 sub _build_balancer {
563 $self->create_balancer(
565 master=>$self->master,
566 %{$self->balancer_args},
570 =head2 _build_write_handler
572 Lazy builder for the L</write_handler> attribute. The default is to set this to
577 sub _build_write_handler {
578 return shift->master;
581 =head2 _build_read_handler
583 Lazy builder for the L</read_handler> attribute. The default is to set this to
588 sub _build_read_handler {
589 return shift->balancer;
592 =head2 around: connect_replicants
594 All calls to connect_replicants needs to have an existing $schema tacked onto
595 top of the args, since L<DBIx::Class::Storage::DBI> needs it, and any
596 L<connect_info|DBIx::Class::Storage::DBI/connect_info>
597 options merged with the master, with replicant opts having higher priority.
601 around connect_replicants => sub {
602 my ($next, $self, @args) = @_;
605 $r = [ $r ] unless reftype $r eq 'ARRAY';
607 $self->throw_exception('coderef replicant connect_info not supported')
608 if ref $r->[0] && reftype $r->[0] eq 'CODE';
610 # any connect_info options?
612 $i++ while $i < @$r && (reftype($r->[$i])||'') ne 'HASH';
615 $r->[$i] = {} unless $r->[$i];
617 # merge if two hashes
618 my @hashes = @$r[$i .. $#{$r}];
620 $self->throw_exception('invalid connect_info options')
621 if (grep { reftype($_) eq 'HASH' } @hashes) != @hashes;
623 $self->throw_exception('too many hashrefs in connect_info')
626 my $merge = Hash::Merge->new('LEFT_PRECEDENT');
627 my %opts = %{ $merge->merge(reverse @hashes) };
630 splice @$r, $i+1, ($#{$r} - $i), ();
632 # make sure master/replicants opts don't clash
633 my %master_opts = %{ $self->_master_connect_info_opts };
634 if (exists $opts{dbh_maker}) {
635 delete @master_opts{qw/dsn user password/};
637 delete $master_opts{dbh_maker};
640 %opts = %{ $merge->merge(\%opts, \%master_opts) };
646 $self->$next($self->schema, @args);
651 Returns an array of all the connected storage backends. The first element
652 in the returned array is the master, and the rest are each of the
659 return grep {defined $_ && blessed $_} (
661 values %{ $self->replicants },
665 =head2 execute_reliably ($coderef, ?@args)
667 Given a coderef, saves the current state of the L</read_handler>, forces it to
668 use reliable storage (e.g. sets it to the master), executes a coderef and then
669 restores the original state.
675 $schema->resultset('User')->create({name=>$name});
676 my $user_rs = $schema->resultset('User')->find({name=>$name});
680 my $user_rs = $schema->storage->execute_reliably($reliably, 'John');
682 Use this when you must be certain of your database state, such as when you just
683 inserted something and need to get a resultset including it, etc.
687 sub execute_reliably {
691 unless( ref $coderef eq 'CODE') {
692 $self->throw_exception('Second argument must be a coderef');
695 ## replace the current read handler for the remainder of the scope
696 local $self->{read_handler} = $self->master;
702 $self->throw_exception("coderef returned an error: $_");
706 =head2 set_reliable_storage
708 Sets the current $schema to be 'reliable', that is all queries, both read and
709 write are sent to the master
713 sub set_reliable_storage {
715 my $schema = $self->schema;
716 my $write_handler = $self->schema->storage->write_handler;
718 $schema->storage->read_handler($write_handler);
721 =head2 set_balanced_storage
723 Sets the current $schema to be use the </balancer> for all reads, while all
724 writes are sent to the master only
728 sub set_balanced_storage {
730 my $schema = $self->schema;
731 my $balanced_handler = $self->schema->storage->balancer;
733 $schema->storage->read_handler($balanced_handler);
738 Check that the master and at least one of the replicants is connected.
745 $self->master->connected &&
746 $self->pool->connected_replicants;
749 =head2 ensure_connected
751 Make sure all the storages are connected.
755 sub ensure_connected {
757 foreach my $source ($self->all_storages) {
758 $source->ensure_connected(@_);
764 Set the limit_dialect for all existing storages
770 foreach my $source ($self->all_storages) {
771 $source->limit_dialect(@_);
773 return $self->master->limit_dialect;
778 Set the quote_char for all existing storages
784 foreach my $source ($self->all_storages) {
785 $source->quote_char(@_);
787 return $self->master->quote_char;
792 Set the name_sep for all existing storages
798 foreach my $source ($self->all_storages) {
799 $source->name_sep(@_);
801 return $self->master->name_sep;
806 Set the schema object for all existing storages
812 foreach my $source ($self->all_storages) {
813 $source->set_schema(@_);
819 set a debug flag across all storages
826 foreach my $source ($self->all_storages) {
830 return $self->master->debug;
841 return $self->master->debugobj(@_);
852 return $self->master->debugfh(@_);
863 return $self->master->debugcb(@_);
868 disconnect everything
874 foreach my $source ($self->all_storages) {
875 $source->disconnect(@_);
881 set cursor class on all storages, or return master's
886 my ($self, $cursor_class) = @_;
889 $_->cursor_class($cursor_class) for $self->all_storages;
891 $self->master->cursor_class;
896 set cursor class on all storages, or return master's, alias for L</cursor_class>
902 my ($self, $cursor_class) = @_;
905 $_->cursor($cursor_class) for $self->all_storages;
907 $self->master->cursor;
912 sets the L<DBIx::Class::Storage::DBI/unsafe> option on all storages or returns
913 master's current setting
921 $_->unsafe(@_) for $self->all_storages;
924 return $self->master->unsafe;
927 =head2 disable_sth_caching
929 sets the L<DBIx::Class::Storage::DBI/disable_sth_caching> option on all storages
930 or returns master's current setting
934 sub disable_sth_caching {
938 $_->disable_sth_caching(@_) for $self->all_storages;
941 return $self->master->disable_sth_caching;
944 =head2 lag_behind_master
946 returns the highest Replicant L<DBIx::Class::Storage::DBI/lag_behind_master>
951 sub lag_behind_master {
954 return max map $_->lag_behind_master, $self->replicants;
957 =head2 is_replicating
959 returns true if all replicants return true for
960 L<DBIx::Class::Storage::DBI/is_replicating>
967 return (grep $_->is_replicating, $self->replicants) == ($self->replicants);
970 =head2 connect_call_datetime_setup
972 calls L<DBIx::Class::Storage::DBI/connect_call_datetime_setup> for all storages
976 sub connect_call_datetime_setup {
978 $_->connect_call_datetime_setup for $self->all_storages;
983 $_->_populate_dbh for $self->all_storages;
988 $_->_connect for $self->all_storages;
993 $_->_rebless for $self->all_storages;
996 sub _determine_driver {
998 $_->_determine_driver for $self->all_storages;
1001 sub _driver_determined {
1005 $_->_driver_determined(@_) for $self->all_storages;
1008 return $self->master->_driver_determined;
1014 $_->_init for $self->all_storages;
1017 sub _run_connection_actions {
1020 $_->_run_connection_actions for $self->all_storages;
1023 sub _do_connection_actions {
1027 $_->_do_connection_actions(@_) for $self->all_storages;
1031 sub connect_call_do_sql {
1033 $_->connect_call_do_sql(@_) for $self->all_storages;
1036 sub disconnect_call_do_sql {
1038 $_->disconnect_call_do_sql(@_) for $self->all_storages;
1041 sub _seems_connected {
1044 return min map $_->_seems_connected, $self->all_storages;
1050 return min map $_->_ping, $self->all_storages;
1053 # not using the normalized_version, because we want to preserve
1054 # version numbers much longer than the conventional xxx.yyyzzz
1055 my $numify_ver = sub {
1057 my @numparts = split /\D+/, $ver;
1058 my $format = '%d.' . (join '', ('%06d') x (@numparts - 1));
1060 return sprintf $format, @numparts;
1065 if (not $self->_dbh_details->{info}) {
1066 $self->_dbh_details->{info} = (
1067 reduce { $a->[0] < $b->[0] ? $a : $b }
1068 map [ $numify_ver->($_->{dbms_version}), $_ ],
1069 map $_->_server_info, $self->all_storages
1073 return $self->next::method;
1076 sub _get_server_version {
1079 return $self->_server_info->{dbms_version};
1084 Due to the fact that replicants can lag behind a master, you must take care to
1085 make sure you use one of the methods to force read queries to a master should
1086 you need realtime data integrity. For example, if you insert a row, and then
1087 immediately re-read it from the database (say, by doing
1088 L<< $result->discard_changes|DBIx::Class::Row/discard_changes >>)
1089 or you insert a row and then immediately build a query that expects that row
1090 to be an item, you should force the master to handle reads. Otherwise, due to
1091 the lag, there is no certainty your data will be in the expected state.
1093 For data integrity, all transactions automatically use the master storage for
1094 all read and write queries. Using a transaction is the preferred and recommended
1095 method to force the master to handle all read queries.
1097 Otherwise, you can force a single query to use the master with the 'force_pool'
1100 my $result = $resultset->search(undef, {force_pool=>'master'})->find($pk);
1102 This attribute will safely be ignored by non replicated storages, so you can use
1103 the same code for both types of systems.
1105 Lastly, you can use the L</execute_reliably> method, which works very much like
1108 For debugging, you can turn replication on/off with the methods L</set_reliable_storage>
1109 and L</set_balanced_storage>, however this operates at a global level and is not
1110 suitable if you have a shared Schema object being used by multiple processes,
1111 such as on a web application server. You can get around this limitation by
1112 using the Schema clone method.
1114 my $new_schema = $schema->clone;
1115 $new_schema->set_reliable_storage;
1117 ## $new_schema will use only the Master storage for all reads/writes while
1118 ## the $schema object will use replicated storage.
1120 =head1 FURTHER QUESTIONS?
1122 Check the list of L<additional DBIC resources|DBIx::Class/GETTING HELP/SUPPORT>.
1124 =head1 COPYRIGHT AND LICENSE
1126 This module is free software L<copyright|DBIx::Class/COPYRIGHT AND LICENSE>
1127 by the L<DBIx::Class (DBIC) authors|DBIx::Class/AUTHORS>. You can
1128 redistribute it and/or modify it under the same terms as the
1129 L<DBIx::Class library|DBIx::Class/COPYRIGHT AND LICENSE>.
1133 __PACKAGE__->meta->make_immutable;