Rails query performance tuning

There are several plugins for this:

https://github.com/dsboulder/query_reviewer

https://github.com/flyerhzm/bullet

and

https://github.com/jeberly/query-analyzer

which works for me ;-)

Mongo, Mysql, Ruby and PHP

Вот тут я сравнивал MongoDB и MySQL в Ruby и результат не сильно меня обрадовал, хотя на MongoDB возлагаю большие надежды.
Я провел некоторые дополнительные бенчмарки.
Скрипты для бенчмарков тут: http://github.com/latrommi/benchmarking
Результаты ниже.
----------------------------------------------------------------
For my English friends:
Some days ago I did some MongoDB benchmark tests.
Here is more extended version.
Benchmark scripts: http://github.com/latrommi/benchmarking
and results below.
----------------------------------------------------------------
Вывод один: это не значит, что MongoDB хуже MySQL, это значит, что для нее ещё не написано нормального, быстрого драйвера для Ruby.

Conclusion: no fast Ruby driver for MongoDB.

MongoDB VS MySQL

Just insert test.

#!/usr/bin/ruby
 

require 'rubygems'
require 'benchmark'
 

require 'mongo'
require 'mysql'
 

# init benchmark cycle
def cycle
    10000.times{ |i|
      doc = {
        "text" => "i am any text and a little tail with me: #{i}",
        "count" => i,
        "coords" => {"x" => 203 + i, "y" => 102 + i}
      }
      

      yield doc
    }
end
 

# init mysql
mysql = Mysql.new('localhost', 'testuser', 'testpass', 'test')
mysql.query("DROP TABLE IF EXISTS `insert_bm`")
mysql.query("DROP TABLE IF EXISTS `insert_bm_coords`")
 

mysql.query("CREATE TABLE `insert_bm_coords`(
  `id` int unsigned not null auto_increment,
  `x` int not null,
  `y` int not null,
  PRIMARY KEY  (`id`)
)")
 

mysql.query("CREATE TABLE `insert_bm`(
  `id` int unsigned not null auto_increment,
  `text` varchar(255) not null,
  `count` int not null,
  `coords_id` int unsigned not null,
  PRIMARY KEY  (`id`)
)")
 

# init mongodb
mongo_db = Mongo::Connection.new.db("ololo_db")
mongo_db.collection("beta").drop
mongo = mongo_db.collection("beta")
 

# do benchmarking
Benchmark.bm {|x|
 

  x.report("mysql:") {
    cycle{|doc|
      mysql.query("INSERT insert_bm_coords(x,y) VALUES(#{doc["coords"]["x"]}, #{doc["coords"]["y"]})")
      mysql.query("INSERT insert_bm(text, count, coords_id) VALUES('#{doc["text"]}', #{doc["count"]}, #{mysql.insert_id()})")
    }
  }
  

  x.report("mongodb:") {
    cycle{|doc|
      mongo.insert(doc)
    }
  }
}
 

Result:

10_000:
   user     system      total        real
mysql:  0.300000   0.150000   0.450000 (  1.763819)
mongodb:  5.690000   0.680000   6.370000 (  7.201129)

1_000_000:
   user     system      total        real
mysql: 30.960000  12.950000  43.910000 (179.318003)
mongodb:569.650000  18.410000 588.060000 (678.889538)

UPDATE: after installing C extension (`gem install mongo_ext`).

10_000:
   user     system      total        real
mysql:  0.280000   0.110000   0.390000 (  1.679100)
mongodb:  3.110000   0.270000   3.380000 (  4.089022)

1_000_000:
   user     system      total        real
mysql: 32.400000  13.030000  45.430000 (179.684186)
mongodb:326.150000  17.000000 343.150000 (423.631974)

I'm confused. Can anybody explain it?

Что-то тут не так, учитывая заявления о скорости MongoDB.
Кто-то может подтвердить или опровергнуть?

UPDATE: here is explanation: http://blog.knopkodav.ru/2009/11/mongo-mysql-ruby-and-php.html

Gentoo + rails + passenger + nginx

I installed ruby by compiling it myself (portage handled ruby doesn't quite do
it for me...). Then I installed the passenger gem 2.2.4. After installing
passenger gem went to the root passenger dir and ran "rake nginx" to get the
helperserver needed. Then I copied the nginx-0.7.61.ebuild and modified it to
include the use flag passenger. Then using the hint from the ebuild attached to
this bug got the full path for the nginx passenger module and put that into my
ebuild.

Then added the passenger use flag to /etc/portage/package.use for nginx and
emerged nginx 0.7.61 with passenger.

Added the config info to /etc/nginx.conf and I'm away everything works very
nicely :)

via http://bugs.gentoo.org/266446

Любителям длинных имён переменных

%username%, помни:
url -- полный путь к странице: http://localhost/blog/2009/07/28/kak-ya-upal-v-rechku
path -- только путь: blog/2009/07/28/kak-ya-upal-v-rechku

И не надо никаких названий переменных, типа: full_path, relative_path, etc.

Converting Drupal tables to RoR from any to any database.

OK, OK, it's fairly not only for Drupal source databases :-)

STEP 1 - init databases connections
Create two new connections in database.yml file like this:

# DATABASE CONVERTING SPECIFIC
# old connection
old:
  adapter: mysql
  database: dbname
  username: username
  password: userpass
  host: localhost

# new connection (may be like your development or production connection)
new:
  adapter: postgresql
  database: dbname
  username: username
  password: userpass
  schema_order: development # for example
  host: localhost

STEP 2 - write rake task
create convert.rake file in your lib/tasks rails application directory

    1 namespace :db do
    2   namespace :convert do
    3     task :old2new => :environment do
    4       print "OK, let's do it now!\n"; STDOUT.flush
    5 
    6       # define universal model for old tables
    7       class OldModel < ActiveRecord::Base
    8       end
    9 
   10       # demonic convertor
   11       def convert(srctable, dstclass)
   12         print "- converting class " + dstclass.to_s + "...\n"; STDOUT.flush
   13 
   14         dstclass.establish_connection(:new)
   15         dstclass.delete_all
   16 
   17         OldModel.set_table_name(srctable)
   18         OldModel.find(:all).each { |old|
   19           new = dstclass.new
   20           new = yield(old, new)
   21           new.save(false)
   22         }
   23       end
   24 
   25       # connect to old dying database
   26       ActiveRecord::Base.establish_connection(:old)
   27 
   28       # CONVERTING
   29 
   30       # users
   31       convert('rctc_users', User) {|old, new|
   32         # prepare each field individually, yeah.
   33         new.id = old.uid
   34         new.login = old.name
   35         new.email = old.mail
   36         new.password = Digest::MD5.hexdigest(old.pass + SALT)
   37         new
   38       }
   39 
   40       # convert your tables below like above
   41       # ...
   42 
   43       print "Wow! Processing complete!\n"
   44     end
   45   end
   46 end

STEP 3 - RUN IT!

$ rake db:convert:old2new

Как починить man, less и другие проблемы tty в OpenVZ

сабж.

# rm /dev/tty
# mknod -m 666 /dev/tty c 5 0


пруфлинк: http://forum.openvz.org/index.php?t=msg&goto=34878&