Design Specs: Database Connectivity

Objectives

Part of the WormBase::API. Needs to be able to connect to multiple databases (multiple instances of the same database in some cases). Should be fault tolerant if a server or database crashes.

Plans

http://bitbucket.org/tharris/wormbase/src/tip/lib/WormBase/API/Role/Service.pm a Moose Role module which defines the common attributes and behavior for different databases.

eg.

'hosts': array reference attribute which has all the available hosts from configuration file.(reading from configuration is not implemented yet!)

reconnect: abstract method for the database connection, fail-over is implementation here: use Moose 'around' method modifier to check on the db handle every time it is used do connect if dbh is undef, host is picked randomly among the pool

  my $host = $self->hosts->[ rand @{$self->hosts} ];

This Role is consumed by each Database Classes(Service module) API/Service/acedb.pm & API/Service/gff.pm

different database will use different modules for connection purpose, here we have

AceDB
use Ace;
Ace->connect
when Ace time out would the dbh becomes undef?
Would the Ace reopen method help a lot? For now I just re-connection again.

GFF database
use Bio::DB::GFF;
Bio::DB::GFF->new(%$gff_args)

Things to think about

persistent connections

Apache::DBI DBI connect_cached DBIx::Class

Also, because Apache::DBI's connection caching never checks to see if the process and/or thread id changed, it will give you an invalid $dbh from a previous process/thread if you use a connection, then thread/fork, then try to re-connect to the same source, which is a Bad Thing.

DBIx::Class already manages its connections for you, and therefore it cannot benefit from Apache::DBI under any scenario. It makes one connection per-process, and keeps that connection persistent, reconnecting only if the connection appears to have died, or if you fork/thread over to another process/thread-id. The only Apache::DBI issue in DBIx::Class is that Apache::DBI will actually thwart DBIx::Class's connection management code, and cause it to use the same (and invalid) connection in a new process, in cases such as (as mentioned above) if you make a DBI connection before forking in a prefork mod_perl server.

Resources

The current connectivity code is part of the API. The API, in turn, is inside of the lib directory of the WormBase mercurial repository:

http://bitbucket.org/tharris/wormbase/src/tip/lib/WormBase/API/Service/

Currently, each database is a "Service". When the API is instantiated, connections are made to each required service. Database handles are cached. When a database handle is needed, the service first checks to see if it is live, if not it refreshes the connection. This would probably be the ideal location to implement failover -- check the current connection, if dead, try a different database server.

• Note that attribute accessors also count as methods for the purposes of satisfying the requirements of a role.

http://bitbucket.org/tharris/wormbase/src/9009075c4c5a/lib/WormBase/API/Role/Service/gff.pm

  around 'dbh' => sub {
    my $orig = shift;
    my $self = shift;
    
    my $species = $self->species;
    
    # Do we already have a dbh? HOW TO TEST THIS WITH HASH REF?
    if ($self->has_dbh) {
	$self->log->debug("     gff-dbh for $species exists and is alive!");
	return $self->$orig;
    } else {
	$self->log->debug("     gff-dbh for $species doesn't exist yet; trying to connect");
	my $dbh = $self->connect($species);
    }
 };

Note also that these Services use Moose "Roles" to define shared methods and variables like connect().

Here's the acedb service:

http://bitbucket.org/tharris/wormbase/src/tip/lib/WormBase/API/Service/acedb.pm

The acedb service "consumes" the generic Service role:

http://bitbucket.org/tharris/wormbase/src/9009075c4c5a/lib/WormBase/API/Role/Service.pm

This nomenclature and structure is a bit confusing and probably needs to be streamlined/refined!!