1 package DBIx::Class::Storage::DBI::Replicated;
5 die('The following modules are required for Replication ' . DBIx::Class::Optional::Dependencies->req_missing_for ('replicated') . "\n" )
6 unless DBIx::Class::Optional::Dependencies->req_ok_for ('replicated');
10 use DBIx::Class::Storage::DBI;
11 use DBIx::Class::Storage::DBI::Replicated::Pool;
12 use DBIx::Class::Storage::DBI::Replicated::Balancer;
13 use DBIx::Class::Storage::DBI::Replicated::Types qw/BalancerClassNamePart DBICSchema DBICStorageDBI/;
14 use MooseX::Types::Moose qw/ClassName HashRef Object/;
15 use Scalar::Util 'reftype';
17 use List::Util qw/min max reduce/;
18 use Context::Preserve 'preserve_context';
21 use namespace::clean -except => 'meta';
25 DBIx::Class::Storage::DBI::Replicated - BETA Replicated database support
29 The Following example shows how to change an existing $schema to a replicated
30 storage type, add some replicated (read-only) databases, and perform reporting
33 You should set the 'storage_type attribute to a replicated type. You should
34 also define your arguments, such as which balancer you want and any arguments
35 that the Pool object should get.
37 my $schema = Schema::Class->clone;
38 $schema->storage_type(['::DBI::Replicated', { balancer_type => '::Random' }]);
39 $schema->connection(...);
41 Next, you need to add in the Replicants. Basically this is an array of
42 arrayrefs, where each arrayref is database connect information. Think of these
43 arguments as what you'd pass to the 'normal' $schema->connect method.
45 $schema->storage->connect_replicants(
46 [$dsn1, $user, $pass, \%opts],
47 [$dsn2, $user, $pass, \%opts],
48 [$dsn3, $user, $pass, \%opts],
51 Now, just use the $schema as you normally would. Automatically all reads will
52 be delegated to the replicants, while writes to the master.
54 $schema->resultset('Source')->search({name=>'etc'});
56 You can force a given query to use a particular storage using the search
57 attribute 'force_pool'. For example:
59 my $rs = $schema->resultset('Source')->search(undef, {force_pool=>'master'});
61 Now $rs will force everything (both reads and writes) to use whatever was setup
62 as the master storage. 'master' is hardcoded to always point to the Master,
63 but you can also use any Replicant name. Please see:
64 L<DBIx::Class::Storage::DBI::Replicated::Pool> and the replicants attribute for more.
66 Also see transactions and L</execute_reliably> for alternative ways to
67 force read traffic to the master. In general, you should wrap your statements
68 in a transaction when you are reading and writing to the same tables at the
69 same time, since your replicants will often lag a bit behind the master.
71 If you have a multi-statement read only transaction you can force it to select
72 a random server in the pool by:
74 my $rs = $schema->resultset('Source')->search( undef,
75 { force_pool => $db->storage->read_handler->next_storage }
80 Warning: This class is marked BETA. This has been running a production
81 website using MySQL native replication as its backend and we have some decent
82 test coverage but the code hasn't yet been stressed by a variety of databases.
83 Individual DBs may have quirks we are not aware of. Please use this in first
84 development and pass along your experiences/bug fixes.
86 This class implements replicated data store for DBI. Currently you can define
87 one master and numerous slave database connections. All write-type queries
88 (INSERT, UPDATE, DELETE and even LAST_INSERT_ID) are routed to master
89 database, all read-type queries (SELECTs) go to the slave database.
91 Basically, any method request that L<DBIx::Class::Storage::DBI> would normally
92 handle gets delegated to one of the two attributes: L</read_handler> or to
93 L</write_handler>. Additionally, some methods need to be distributed
94 to all existing storages. This way our storage class is a drop in replacement
95 for L<DBIx::Class::Storage::DBI>.
97 Read traffic is spread across the replicants (slaves) occurring to a user
98 selected algorithm. The default algorithm is random weighted.
102 The consistency between master and replicants is database specific. The Pool
103 gives you a method to validate its replicants, removing and replacing them
104 when they fail/pass predefined criteria. Please make careful use of the ways
105 to force a query to run against Master when needed.
109 Replicated Storage has additional requirements not currently part of
110 L<DBIx::Class>. See L<DBIx::Class::Optional::Dependencies> for more details.
114 This class defines the following attributes.
118 The underlying L<DBIx::Class::Schema> object this storage is attaching
131 Contains the classname which will instantiate the L</pool> object. Defaults
132 to: L<DBIx::Class::Storage::DBI::Replicated::Pool>.
139 default=>'DBIx::Class::Storage::DBI::Replicated::Pool',
141 'create_pool' => 'new',
147 Contains a hashref of initialized information to pass to the Balancer object.
148 See L<DBIx::Class::Storage::DBI::Replicated::Pool> for available arguments.
162 The replication pool requires a balance class to provider the methods for
163 choose how to spread the query load across each replicant in the pool.
167 has 'balancer_type' => (
169 isa=>BalancerClassNamePart,
172 default=> 'DBIx::Class::Storage::DBI::Replicated::Balancer::First',
174 'create_balancer' => 'new',
180 Contains a hashref of initialized information to pass to the Balancer object.
181 See L<DBIx::Class::Storage::DBI::Replicated::Balancer> for available arguments.
185 has 'balancer_args' => (
195 Is a L<DBIx::Class::Storage::DBI::Replicated::Pool> or derived class. This is a
196 container class for one or more replicated databases.
202 isa=>'DBIx::Class::Storage::DBI::Replicated::Pool',
213 Is a L<DBIx::Class::Storage::DBI::Replicated::Balancer> or derived class. This
214 is a class that takes a pool (L<DBIx::Class::Storage::DBI::Replicated::Pool>)
220 isa=>'DBIx::Class::Storage::DBI::Replicated::Balancer',
222 handles=>[qw/auto_validate_every/],
227 The master defines the canonical state for a pool of connected databases. All
228 the replicants are expected to match this databases state. Thus, in a classic
229 Master / Slaves distributed system, all the slaves are expected to replicate
230 the Master's state as quick as possible. This is the only database in the
231 pool of databases that is allowed to handle write traffic.
241 =head1 ATTRIBUTES IMPLEMENTING THE DBIx::Storage::DBI INTERFACE
243 The following methods are delegated all the methods required for the
244 L<DBIx::Class::Storage::DBI> interface.
248 my $method_dispatch = {
260 deployment_statements
263 build_datetime_parser
278 with_deferred_fk_checks
286 relname_to_table_alias
288 _default_dbi_connect_attributes
290 _dbic_connect_attributes
296 bind_attribute_by_data_type
300 _dbh_execute_for_fetch
302 _dbh_execute_inserts_with_no_binds
303 _select_args_to_query
306 _normalize_connect_info
320 /, Class::MOP::Class->initialize('DBIx::Class::Storage::DBIHacks')->get_method_list ],
325 _dbh_columns_info_for
328 unimplemented => [qw/
329 _arm_global_destructor
332 get_use_dbms_capability
333 set_use_dbms_capability
339 _determine_connector_driver
340 _extract_driver_from_connect_info
342 _warn_undetermined_driver
349 _perform_autoinc_retrieval
350 _autoinc_supplied_for_op
363 # the capability framework
364 # not sure if CMOP->initialize does evil things to DBIC::S::DBI, fix if a problem
366 { $_ =~ /^ _ (?: use | supports | determine_supports ) _ /x }
367 ( Class::MOP::Class->initialize('DBIx::Class::Storage::DBI')->get_all_method_names )
371 if (DBIx::Class::_ENV_::DBICTEST) {
374 for my $type (keys %$method_dispatch) {
375 for (@{$method_dispatch->{$type}}) {
376 push @{$seen->{$_}}, $type;
380 if (my @dupes = grep { @{$seen->{$_}} > 1 } keys %$seen) {
382 'The following methods show up multiple times in ::Storage::DBI::Replicated handlers:',
383 (map { "$_: " . (join ', ', @{$seen->{$_}}) } sort @dupes),
388 if (my @cant = grep { ! DBIx::Class::Storage::DBI->can($_) } keys %$seen) {
390 '::Storage::DBI::Replicated specifies handling of the following *NON EXISTING* ::Storage::DBI methods:',
397 for my $method (@{$method_dispatch->{unimplemented}}) {
398 __PACKAGE__->meta->add_method($method, sub {
400 $self->throw_exception("$method() must not be called on ".(blessed $self).' objects');
406 Defines an object that implements the read side of L<DBIx::Class::Storage::DBI>.
410 has 'read_handler' => (
414 handles=>$method_dispatch->{reader},
419 Defines an object that implements the write side of L<DBIx::Class::Storage::DBI>,
420 as well as methods that don't write or read that can be called on only one
421 storage, methods that return a C<$dbh>, and any methods that don't make sense to
426 has 'write_handler' => (
430 handles=>$method_dispatch->{writer},
435 has _master_connect_info_opts =>
436 (is => 'rw', isa => HashRef, default => sub { {} });
438 =head2 around: connect_info
440 Preserves master's C<connect_info> options (for merging with replicants.)
441 Also sets any Replicated-related options from connect_info, such as
442 C<pool_type>, C<pool_args>, C<balancer_type> and C<balancer_args>.
446 around connect_info => sub {
447 my ($next, $self, $info, @extra) = @_;
449 $self->throw_exception(
450 'connect_info can not be retrieved from a replicated storage - '
451 . 'accessor must be called on a specific pool instance'
452 ) unless defined $info;
454 my $merge = Hash::Merge->new('LEFT_PRECEDENT');
457 for my $arg (@$info) {
458 next unless (reftype($arg)||'') eq 'HASH';
459 %opts = %{ $merge->merge($arg, \%opts) };
463 if (@opts{qw/pool_type pool_args/}) {
464 $self->pool_type(delete $opts{pool_type})
468 $merge->merge((delete $opts{pool_args} || {}), $self->pool_args)
471 ## Since we possibly changed the pool_args, we need to clear the current
472 ## pool object so that next time it is used it will be rebuilt.
476 if (@opts{qw/balancer_type balancer_args/}) {
477 $self->balancer_type(delete $opts{balancer_type})
478 if $opts{balancer_type};
480 $self->balancer_args(
481 $merge->merge((delete $opts{balancer_args} || {}), $self->balancer_args)
484 $self->balancer($self->_build_balancer)
488 $self->_master_connect_info_opts(\%opts);
490 return preserve_context {
491 $self->$next($info, @extra);
493 # Make sure master is blessed into the correct class and apply role to it.
494 my $master = $self->master;
495 $master->_determine_driver;
496 Moose::Meta::Class->initialize(ref $master);
498 DBIx::Class::Storage::DBI::Replicated::WithDSN->meta->apply($master);
500 # link pool back to master
501 $self->pool->master($master);
507 This class defines the following methods.
511 L<DBIx::Class::Schema> when instantiating its storage passed itself as the
512 first argument. So we need to massage the arguments a bit so that all the
513 bits get put into the correct places.
518 my ($class, $schema, $storage_type_args, @args) = @_;
529 Lazy builder for the L</master> attribute.
535 my $master = DBIx::Class::Storage::DBI->new($self->schema);
541 Lazy builder for the L</pool> attribute.
547 $self->create_pool(%{$self->pool_args});
550 =head2 _build_balancer
552 Lazy builder for the L</balancer> attribute. This takes a Pool object so that
553 the balancer knows which pool it's balancing.
557 sub _build_balancer {
559 $self->create_balancer(
561 master=>$self->master,
562 %{$self->balancer_args},
566 =head2 _build_write_handler
568 Lazy builder for the L</write_handler> attribute. The default is to set this to
573 sub _build_write_handler {
574 return shift->master;
577 =head2 _build_read_handler
579 Lazy builder for the L</read_handler> attribute. The default is to set this to
584 sub _build_read_handler {
585 return shift->balancer;
588 =head2 around: connect_replicants
590 All calls to connect_replicants needs to have an existing $schema tacked onto
591 top of the args, since L<DBIx::Class::Storage::DBI> needs it, and any
592 L<connect_info|DBIx::Class::Storage::DBI/connect_info>
593 options merged with the master, with replicant opts having higher priority.
597 around connect_replicants => sub {
598 my ($next, $self, @args) = @_;
601 $r = [ $r ] unless reftype $r eq 'ARRAY';
603 $self->throw_exception('coderef replicant connect_info not supported')
604 if ref $r->[0] && reftype $r->[0] eq 'CODE';
606 # any connect_info options?
608 $i++ while $i < @$r && (reftype($r->[$i])||'') ne 'HASH';
611 $r->[$i] = {} unless $r->[$i];
613 # merge if two hashes
614 my @hashes = @$r[$i .. $#{$r}];
616 $self->throw_exception('invalid connect_info options')
617 if (grep { reftype($_) eq 'HASH' } @hashes) != @hashes;
619 $self->throw_exception('too many hashrefs in connect_info')
622 my $merge = Hash::Merge->new('LEFT_PRECEDENT');
623 my %opts = %{ $merge->merge(reverse @hashes) };
626 splice @$r, $i+1, ($#{$r} - $i), ();
628 # make sure master/replicants opts don't clash
629 my %master_opts = %{ $self->_master_connect_info_opts };
630 if (exists $opts{dbh_maker}) {
631 delete @master_opts{qw/dsn user password/};
633 delete $master_opts{dbh_maker};
636 %opts = %{ $merge->merge(\%opts, \%master_opts) };
642 $self->$next($self->schema, @args);
647 Returns an array of all the connected storage backends. The first element
648 in the returned array is the master, and the rest are each of the
655 return grep {defined $_ && blessed $_} (
657 values %{ $self->replicants },
661 =head2 execute_reliably ($coderef, ?@args)
663 Given a coderef, saves the current state of the L</read_handler>, forces it to
664 use reliable storage (e.g. sets it to the master), executes a coderef and then
665 restores the original state.
671 $schema->resultset('User')->create({name=>$name});
672 my $user_rs = $schema->resultset('User')->find({name=>$name});
676 my $user_rs = $schema->storage->execute_reliably($reliably, 'John');
678 Use this when you must be certain of your database state, such as when you just
679 inserted something and need to get a resultset including it, etc.
683 sub execute_reliably {
687 unless( ref $coderef eq 'CODE') {
688 $self->throw_exception('Second argument must be a coderef');
691 ## replace the current read handler for the remainder of the scope
692 local $self->{read_handler} = $self->master;
698 $self->throw_exception("coderef returned an error: $_");
702 =head2 set_reliable_storage
704 Sets the current $schema to be 'reliable', that is all queries, both read and
705 write are sent to the master
709 sub set_reliable_storage {
711 my $schema = $self->schema;
712 my $write_handler = $self->schema->storage->write_handler;
714 $schema->storage->read_handler($write_handler);
717 =head2 set_balanced_storage
719 Sets the current $schema to be use the </balancer> for all reads, while all
720 writes are sent to the master only
724 sub set_balanced_storage {
726 my $schema = $self->schema;
727 my $balanced_handler = $self->schema->storage->balancer;
729 $schema->storage->read_handler($balanced_handler);
734 Check that the master and at least one of the replicants is connected.
741 $self->master->connected &&
742 $self->pool->connected_replicants;
745 =head2 ensure_connected
747 Make sure all the storages are connected.
751 sub ensure_connected {
753 foreach my $source ($self->all_storages) {
754 $source->ensure_connected(@_);
760 Set the limit_dialect for all existing storages
766 foreach my $source ($self->all_storages) {
767 $source->limit_dialect(@_);
769 return $self->master->limit_dialect;
774 Set the quote_char for all existing storages
780 foreach my $source ($self->all_storages) {
781 $source->quote_char(@_);
783 return $self->master->quote_char;
788 Set the name_sep for all existing storages
794 foreach my $source ($self->all_storages) {
795 $source->name_sep(@_);
797 return $self->master->name_sep;
802 Set the schema object for all existing storages
808 foreach my $source ($self->all_storages) {
809 $source->set_schema(@_);
815 set a debug flag across all storages
822 foreach my $source ($self->all_storages) {
826 return $self->master->debug;
837 return $self->master->debugobj(@_);
848 return $self->master->debugfh(@_);
859 return $self->master->debugcb(@_);
864 disconnect everything
870 foreach my $source ($self->all_storages) {
871 $source->disconnect(@_);
877 set cursor class on all storages, or return master's
882 my ($self, $cursor_class) = @_;
885 $_->cursor_class($cursor_class) for $self->all_storages;
887 $self->master->cursor_class;
892 set cursor class on all storages, or return master's, alias for L</cursor_class>
898 my ($self, $cursor_class) = @_;
901 $_->cursor($cursor_class) for $self->all_storages;
903 $self->master->cursor;
908 sets the L<DBIx::Class::Storage::DBI/unsafe> option on all storages or returns
909 master's current setting
917 $_->unsafe(@_) for $self->all_storages;
920 return $self->master->unsafe;
923 =head2 disable_sth_caching
925 sets the L<DBIx::Class::Storage::DBI/disable_sth_caching> option on all storages
926 or returns master's current setting
930 sub disable_sth_caching {
934 $_->disable_sth_caching(@_) for $self->all_storages;
937 return $self->master->disable_sth_caching;
940 =head2 lag_behind_master
942 returns the highest Replicant L<DBIx::Class::Storage::DBI/lag_behind_master>
947 sub lag_behind_master {
950 return max map $_->lag_behind_master, $self->replicants;
953 =head2 is_replicating
955 returns true if all replicants return true for
956 L<DBIx::Class::Storage::DBI/is_replicating>
963 return (grep $_->is_replicating, $self->replicants) == ($self->replicants);
966 =head2 connect_call_datetime_setup
968 calls L<DBIx::Class::Storage::DBI/connect_call_datetime_setup> for all storages
972 sub connect_call_datetime_setup {
974 $_->connect_call_datetime_setup for $self->all_storages;
979 $_->_populate_dbh for $self->all_storages;
984 $_->_connect for $self->all_storages;
989 $_->_rebless for $self->all_storages;
992 sub _determine_driver {
994 $_->_determine_driver for $self->all_storages;
997 sub _driver_determined {
1001 $_->_driver_determined(@_) for $self->all_storages;
1004 return $self->master->_driver_determined;
1010 $_->_init for $self->all_storages;
1013 sub _run_connection_actions {
1016 $_->_run_connection_actions for $self->all_storages;
1019 sub _do_connection_actions {
1023 $_->_do_connection_actions(@_) for $self->all_storages;
1027 sub connect_call_do_sql {
1029 $_->connect_call_do_sql(@_) for $self->all_storages;
1032 sub disconnect_call_do_sql {
1034 $_->disconnect_call_do_sql(@_) for $self->all_storages;
1037 sub _seems_connected {
1040 return min map $_->_seems_connected, $self->all_storages;
1046 return min map $_->_ping, $self->all_storages;
1049 # not using the normalized_version, because we want to preserve
1050 # version numbers much longer than the conventional xxx.yyyzzz
1051 my $numify_ver = sub {
1053 my @numparts = split /\D+/, $ver;
1054 my $format = '%d.' . (join '', ('%06d') x (@numparts - 1));
1056 return sprintf $format, @numparts;
1061 if (not $self->_dbh_details->{info}) {
1062 $self->_dbh_details->{info} = (
1063 reduce { $a->[0] < $b->[0] ? $a : $b }
1064 map [ $numify_ver->($_->{dbms_version}), $_ ],
1065 map $_->_server_info, $self->all_storages
1069 return $self->next::method;
1072 sub _get_server_version {
1075 return $self->_server_info->{dbms_version};
1080 Due to the fact that replicants can lag behind a master, you must take care to
1081 make sure you use one of the methods to force read queries to a master should
1082 you need realtime data integrity. For example, if you insert a row, and then
1083 immediately re-read it from the database (say, by doing
1084 L<< $result->discard_changes|DBIx::Class::Row/discard_changes >>)
1085 or you insert a row and then immediately build a query that expects that row
1086 to be an item, you should force the master to handle reads. Otherwise, due to
1087 the lag, there is no certainty your data will be in the expected state.
1089 For data integrity, all transactions automatically use the master storage for
1090 all read and write queries. Using a transaction is the preferred and recommended
1091 method to force the master to handle all read queries.
1093 Otherwise, you can force a single query to use the master with the 'force_pool'
1096 my $result = $resultset->search(undef, {force_pool=>'master'})->find($pk);
1098 This attribute will safely be ignored by non replicated storages, so you can use
1099 the same code for both types of systems.
1101 Lastly, you can use the L</execute_reliably> method, which works very much like
1104 For debugging, you can turn replication on/off with the methods L</set_reliable_storage>
1105 and L</set_balanced_storage>, however this operates at a global level and is not
1106 suitable if you have a shared Schema object being used by multiple processes,
1107 such as on a web application server. You can get around this limitation by
1108 using the Schema clone method.
1110 my $new_schema = $schema->clone;
1111 $new_schema->set_reliable_storage;
1113 ## $new_schema will use only the Master storage for all reads/writes while
1114 ## the $schema object will use replicated storage.
1116 =head1 FURTHER QUESTIONS?
1118 Check the list of L<additional DBIC resources|DBIx::Class/GETTING HELP/SUPPORT>.
1120 =head1 COPYRIGHT AND LICENSE
1122 This module is free software L<copyright|DBIx::Class/COPYRIGHT AND LICENSE>
1123 by the L<DBIx::Class (DBIC) authors|DBIx::Class/AUTHORS>. You can
1124 redistribute it and/or modify it under the same terms as the
1125 L<DBIx::Class library|DBIx::Class/COPYRIGHT AND LICENSE>.
1129 __PACKAGE__->meta->make_immutable;