Class: Pgtk::Pool

Inherits:
Object
  • Object
show all
Defined in:
lib/pgtk/pool.rb

Overview

Pool provides a connection pool for PostgreSQL database connections. It manages a fixed number of connections to optimize performance and resource usage while providing a simple interface for database operations.

The Pool class handles connection lifecycle, reconnects on errors, and provides transaction support. It’s the core class for interacting with a PostgreSQL database in this library.

Basic usage:

# Create a wire that knows how to connect to PostgreSQL
wire = Pgtk::Wire::Direct.new(
  host: 'localhost',
  port: 5432,
  dbname: 'mydatabase',
  user: 'postgres',
  password: 'secret'
)

# Create and start a connection pool with 4 connections
pool = Pgtk::Pool.new(wire, max: 4)
pool.start!

# Execute a simple query
pool.exec('SELECT * FROM users')

# Execute a parameterized query
pool.exec('SELECT * FROM users WHERE email = $1', ['[email protected]'])

# Use transactions for multiple operations
pool.transaction do |t|
  t.exec('UPDATE accounts SET balance = balance - $1 WHERE id = $2', [100, 42])
  t.exec('UPDATE accounts SET balance = balance + $1 WHERE id = $2', [100, 43])
end
Author

Yegor Bugayenko ([email protected])

Copyright

Copyright © 2019-2025 Yegor Bugayenko

License

MIT

Defined Under Namespace

Classes: IterableQueue, Txn

Instance Method Summary collapse

Constructor Details

#initialize(wire, max: 8, log: Loog::NULL) ⇒ Pool

Constructor.

Parameters:

  • wire (Pgtk::Wire) —

    The wire

  • max (Integer) (defaults to: 8) —

    Total amount of PostgreSQL connections in the pool

  • log (Object) (defaults to: Loog::NULL) —

    The log



57
58
59
60
61
62
63
 
# File 'lib/pgtk/pool.rb', line 57

def initialize(wire, max: 8, log: Loog::NULL)
  @wire = wire
  @max = max
  @log = log
  @pool = IterableQueue.new(max)
  @started = false
end

Instance Method Details

#dump ⇒ String

Get as much details about it as possible.

Returns:

  • (String) —

    Summary of inner state



75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
 
# File 'lib/pgtk/pool.rb', line 75

def dump
  [
    'Pgtk::Pool',
    "  Pgtk version: #{Pgtk::VERSION}",
    "  PgSQL version: #{version}",
    "  #{@pool.size} connections:",
    @pool.map do |c|
      [
        '    ',
        "##{c.backend_pid}",
        case c.pipeline_status
        when PG::Constants::PQ_PIPELINE_ON
          'ON'
        when PG::Constants::PQ_PIPELINE_OFF
          'OFF'
        when PG::Constants::PQ_PIPELINE_ABORTED
          'ABORTED'
        else
          "pipeline_status=#{c.pipeline_status}"
        end,
        case c.status
        when PG::Constants::CONNECTION_OK
          'OK'
        when PG::Constants::CONNECTION_BAD
          'BAD'
        else
          "status=#{c.status}"
        end,
        case c.transaction_status
        when PG::Constants::PQTRANS_IDLE
          'IDLE'
        when PG::Constants::PQTRANS_ACTIVE
          'ACTIVE'
        when PG::Constants::PQTRANS_INTRANS
          'INTRANS'
        when PG::Constants::PQTRANS_INERROR
          'INERROR'
        when PG::Constants::PQTRANS_UNKNOWN
          'UNKNOWN'
        else
          "transaction_status=#{c.transaction_status}"
        end
      ].join(' ')
    rescue PG::ConnectionBad => e
      e.message
    end
  ].flatten.join("\n")
end

#exec(query, args = [], result = 0) {|Hash| ... } ⇒ Object

Make a query and return the result as an array of hashes. For example, in order to fetch the list of all books belonging to the user:

books = pool.exec('SELECT * FROM book WHERE owner = $1', ['yegor256'])
books.each do |row|
  puts 'ID: ' + row['id'].to_i
  puts 'Created: ' + Time.parse(row['created'])
  puts 'Title: ' + row['title']
end

All values in the retrieved hash are strings. No matter what types of of data you have in the database, you get strings here. It’s your job to convert them to the type you need.

In order to insert a new row (pay attention to the RETURNING clause at the end of the SQL query):

id = pool.exec(
  'INSERT INTO book (owner, title) VALUES ($1, $2) RETURNING id',
  ['yegor256', 'Elegant Objects']
)[0]['id'].to_i

You can also pass a block to this method, if you want to get an instance of PG::Result instead of an array of hashes:

pool.exec('SELECT * FROM book WHERE owner = $1', ['yegor256']) do |res|
  res.each do |row|
    puts 'ID: ' + row['id'].to_i
    puts 'Title: ' + row['title']
  end
end

When the query is too long it’s convenient to use an array to specify it:

pool.exec(
  [
    'SELECT * FROM book',
    'LEFT JOIN user ON user.id = book.owner',
    'WHERE user.login = $1 AND book.title = $2'
  ],
  ['yegor256', 'Elegant Objects']
)

More details about exec_params, which is called here, you can find here: www.rubydoc.info/gems/pg/0.17.1/PG%2FConnection:exec_params

Parameters:

  • query (String) —

    The SQL query with params inside (possibly)

  • args (Array) (defaults to: []) —

    List of arguments

  • result (Integer) (defaults to: 0) —

    Should be 0 for text results, 1 for binary

Yields:

  • (Hash) —

    Rows



189
190
191
192
193
194
195
196
197
198
 
# File 'lib/pgtk/pool.rb', line 189

def exec(query, args = [], result = 0, &block)
  connect do |c|
    t = Txn.new(c, @log)
    if block_given?
      t.exec(query, args, result, &block)
    else
      t.exec(query, args, result)
    end
  end
end

#start! ⇒ Object

Start it with a fixed number of connections. The amount of connections is specified in max argument and should be big enough to handle the amount of parallel connections you may have to the database. However, keep in mind that not all servers will allow you to have many connections open at the same time. For example, Heroku free PostgreSQL database allows only one connection open.



130
131
132
133
134
135
136
137
 
# File 'lib/pgtk/pool.rb', line 130

def start!
  return if @started
  @max.times do
    @pool << @wire.connection
  end
  @started = true
  @log.debug("PostgreSQL pool started with #{@max} connections")
end

#transaction ⇒ Object

Run a transaction. The block has to be provided. It will receive a temporary object, which implements method exec, which works exactly like the method exec of class Pool, for example:

pgsql.transaction do |t|
  t.exec('DELETE FROM user WHERE id = $1', [id])
  t.exec('INSERT INTO user (name) VALUES ($1)', [name])
end



208
209
210
211
212
213
214
215
216
217
218
219
220
221
 
# File 'lib/pgtk/pool.rb', line 208

def transaction
  connect do |c|
    t = Txn.new(c, @log)
    t.exec('START TRANSACTION')
    begin
      r = yield t
      t.exec('COMMIT')
      r
    rescue StandardError => e
      t.exec('ROLLBACK')
      raise e
    end
  end
end

#version ⇒ String

Get the version of PostgreSQL server.

Returns:

  • (String) —

    Version of PostgreSQL server



68
69
70
 
# File 'lib/pgtk/pool.rb', line 68

def version
  @version ||= exec('SHOW server_version')[0]['server_version'].split[0]
end