Trinidad allows you to run Rails and/or Rack applications within an embedded Tomcat container. Apache Tomcat (formerly also Jakarta Tomcat) is an open source web server and Servlet container with a long history that dates back to the previous millenia.
Trinidad's goals with bringing Tomcat into JRuby land are mostly the following :
JRuby::Rack, thus even
if you do some Java integration or use it's Rack Servlet extensions you're
still able to migrate to a standalone Tomcat or any other Servlet containerResque and Delayed::Job
$ jruby -S gem install trinidad
NOTE: please use 1.5.0.B2 gem install trinidad --pre, esp. on JRuby 9K,
as Trinidad 1.4 will no longer receive Tomcat (7.0.x) security updates.
Trinidad 1.4 requires (and supports) JRuby 1.6.8 or later (latest 1.7.x recommended).
$ cd a-rails-app
$ jruby -S trinidad
If you use Bundler, you might want to add Trinidad to your Gemfile :
gem 'trinidad', :require => nil
However this is not absolutely necessary, you might simply gem install trinidad
and than run trinidad - keep in mind a server is not an application dependency.
Trinidad supports the same Rails version as the JRuby-Rack it founds on (or is specified/locked in your Gemfile), which are 4.x, 3.x and even 2.3 for JRuby-Rack 1.1.x (and the up coming 1.2). Merb is not supported.
$ trinidad
or if you prefer to use the Rack handler (e.g. for development) use :
$ rails s trinidad
Please note all configuration options will work using the Rack handler mode, you
should usually only use it for development/tests and run trinidad on production.
$ ruby app.rb -s Trinidad
or configure your application to always use Trinidad :
require 'sinatra'
require 'trinidad'
configure do
set :server, :trinidad
end
Trinidad auto-detects a plain-old Rack application (if there's a config.ru) :
$ trinidad
You can as well pass the server name to rackup to start the Rack handler :
$ rackup -s trinidad
Or you can set Trinidad as the default server in your config.ru file :
#\ -s trinidad
Trinidad solves the Rack "chicken-egg" problem when booting from a rackup file the same way as JRuby-Rack (since it boots all applications), that is :
# rack.version: ~>1.4.0 (or the latest installed gem is used)NOTE: We recommend to use the plain trinidad mode for running apps
(in production), since it supports runtime pooling while the "rackup" mode does
not, it also provides you with better Java integration possibilities.
Also note that Trinidad does not mimic JRuby-Rack's (1.1.x) backwards compatible behavior of starting a pool for Rails but booting a thread-safe runtime for plain Rack applications by default. Runtime pooling is the default with Trinidad 1.4 and stays the same no matter the type of the application. This has changed in Trinidad 1.5 and it assumes thread-safe applications by default.
All major rack versions (< 2.0) are expected to be working fine with Trinidad.
Trinidad allows you to configure parameters from the command line, the following
is a list of the currently supported options (try trinidad -h):
* -d, --dir ROOT_DIR => web application root directory
* -e, --env ENVIRONMENT => rack (rails) environment
* --rackup [RACKUP_FILE] => rackup configuration file
* --public PUBLIC_DIR => web application public root
* -c, --context CONTEXT => application context path
* --monitor MONITOR_FILE => monitor for application re-deploys
* -t, --threadsafe => force thread-safe mode (use single runtime)
* --runtimes MIN:MAX => use given number of min/max jruby runtimes
* -f, --config [CONFIG_FILE] => configuration file
* --address ADDRESS => host address
* -p, --port PORT => port to bind to
* -s, --ssl [SSL_PORT] => enable secure socket layout
* -a, --ajp [AJP_PORT] => enable the AJP web protocol
* --java_lib LIB_DIR => contains .jar files used by the app
* --java_classes CLASSES_DIR => contains java classes used by the app
* -l, --load EXTENSION_NAMES => load options for extensions
* --apps_base APPS_BASE_DIR => set applications base directory
* -g, --log LEVEL => set logging level
You can also specify a default web.xml to configure your web application.
By default the server tries to load the file config/web.xml but you can change
the path by adding the option default_web_xml within your configuration file.
The server can be configured from a .yml file. By default, if a file is not specified, the server tries to load config/trinidad.yml. Within this file you can specify options available on the command line and tune server settings or configure multiple applications to be hosted on the server.
Advanced configuration options are explained in the wiki: http://wiki.github.com/trinidad/trinidad/advanced-configuration
$ jruby -S trinidad --config my_trinidad.yml
---
port: 4242
address: 0.0.0.0
As an alternative to the config/trinidad.yml file, a .rb configuration file might be used to setup Trinidad. It follows the same convention as the YAML configuration - the file config/trinidad.rb is loaded by default if exists.
Trinidad.configure do |config|
config.port = 4242
config.address = '0.0.0.0'
#config[:custom] = 'custom'
end
As you might notice on your first trinidad the server uses standard output :
kares@theborg:~/workspace/trinidad/MegaUpload$ trinidad -p 8000 -e staging
Initializing ProtocolHandler ["http-bio-8000"]
Starting Servlet Engine: Apache Tomcat/7.0.28
Starting ProtocolHandler ["http-bio-8000"]
Context with name [/] has started rolling
Context with name [/] has completed rolling
It also prints warnings and error messages on error output, while application
specific log messages (e.g. logs from Rails.logger) always go into the expected
file location at log/{environment}.log.
Application logging performs daily file rolling out of the box and only prints messages to the console while it runs in development mode, that means you won't see any application specific output on the console say in production !
Please note that these logging details as well as the logging format will be configurable with trinidad.yml/.rb within the next 1.4.x release.
If you plan to use a slice of Java with your JRuby and require a logger, consider
using ServletContext#log. By default it is setup in a way that logging with
ServletContext ends up in the same location as the Rails log.
If this is not enough you can still configure a Java logging library e.g. SLF4J,
just make sure you tell Trinidad to use it as well, if needed, using the
jruby.rack.logging context parameter in web.xml.
For slightly advanced (and "dirty" XML :)) application configuration Trinidad also supports the exact same context.xml format as Tomcat. Each web app is represented as a context instance and might be configured as such. You do not need to repeat configuring the same parameters you have already setup with the Trinidad configuration. This is meant to be mostly for those familiar with Tomcat internals. Currently the application's context.xml is expected to be located on the class-path under your [classes]/META-INF directory.
Context Doc: http://tomcat.apache.org/tomcat-7.0-doc/config/context.html
Trinidad uses Tomcat's built-in capabilities to server your public files. We do recommend compiling assets up front and disabling the asset server (in production) if you're using the asset pipeline in a Rails application. If you do not put a web-server such as Apache in front of Trinidad you might want to configure the resource caching (on by default for env != development) for maximum performance e.g. by default it's configured as follows :
---
public:
root: public # same as the above "public: public" setting
cached: true # enable (in-memory) asset caching on for env != 'development'
cache_ttl: 5000 # cache TTL in millis (might want to increase this)
cache_max_size: 10240 # the maximum cache size in kB
cache_object_max_size: 512 # max size for a cached object (asset) in kB
#aliases: # allows to "link" other directories into the public root e.g. :
#/home: /var/local/www
Note that this configuration applies to (server-side) resource caching on top of the "public" file-system. You do not need to worry about client side caching, it is handled out of the box with ETag and Last-Modified headers being set.
You might also "mount" file-system directories as aliases to your resources root to be served by your application (as if they were in the public folder).
NOTE: In development mode if you ever happen to rake assets:precompile
make sure to remove your public/assets directory later, otherwise requests
such as /assets/application.js?body=1.0 might not hit the Rails runtime.
Trinidad supports monitoring a file to reload applications, when the file
tmp/restart.txt is updated (e.g. touch tmp/restart.txt on Unix or
type nul >>tmp\restart.txt & copy /b tmp\restart.txt +,, on Windows),
the server reloads the application the monitor file belongs to.
This monitor file can be customized with the monitor configuration option.
Since version 1.4.0 Trinidad supports 2 reload strategies :
restart (default) synchronous reloading. This strategy pauses incoming requests while it reloads the application and then serves them once ready (or timeouts if it takes too long). It is the default strategy since 1.4.0 due it's more predictable memory requirements.
rolling a.k.a. "zero-downtime" (asynchronous) reloading strategy similar to Passenger's rolling reloads. This has been the default since 1.1.0 up till the 1.3.x line. If you use this you should account that your JVM memory requirements might increase quite a lot (esp. if you reload under heavy loads) since requests are being served while there's another version of the application being loaded.
NOTE: due the way class-loaders where setup internally, Trinidad might have failed releasing memory with reloads. This has been fixed in 1.5.0 please consider updating, it is meant to be backwards compatible.
If you're on Java 6 you will likely need to tune your JAVA_OPTS / JRUBY_OPTS for the JVM to do class unloading (consult the wiki for more information) :
JRUBY_OPTS="$JRUBY_OPTS -J-XX:+UseConcMarkSweepGC -J-XX:+CMSClassUnloadingEnabled"
Configure the reload strategy per web application or globally e.g. :
---
port: 8080
environment: production
reload_strategy: rolling
It's possible to use Trinidad with multiple hosts and load the applications under them automatically. A (virtual) host represents an association of a network name (such as "www.example.com" with the particular server on which Tomcat is running. Please remember that each host must have its applications in a different directory. You can find out more at Tomcat's documentation.
Trinidad.configure do |config|
config.hosts = {
# applications path (host app base directory) => host names
# (first one is the "main" host name, other ones are aliases)
'/var/www/local/apps' => ['localhost', '127.0.0.1'],
'/home/trinidad/apps' => 'appshost'
# NOTE: by default a (default) 'localhost' host is setup
}
end
Detailed host configuration is also possible using supported host options :
---
port: 8080
hosts:
localhost:
app_base: /home/trinidad/apps
auto_deploy: false
unpackWARs: true
If applications are configured via the web_apps section, the host for each
application can be added with the host (or hosts) key, if a specified host
does not exists (e.g. not configured or not "localhost") it will be created.
If several applications belong to the same host, they are expected to reside
under the same parent directory e.g. :
Trinidad.configure do |config|
config.web_apps = {
:mock1 => {
:root_dir => 'rails_apps/mock1',
:host => ['rails.virtual.host', 'rails.host']
},
:mock2 => {
:root_dir => 'rails_apps/mock2',
:host => 'rails.virtual.host'
},
:mock3 => {
:root_dir => 'rack_apps/mock3',
:host => ['rack.virtual.host', 'rack.host']
}
}
end
Trinidad allows to extend itself with more (not just Tomcat) features using extensions, they're essentially components hooked into Tomcat's life-cycle. Here is a list of the available extensions that are "officially supported" :
You can find further information on how to write extensions in the wiki.
Copyright (c) 2016 Team Trinidad. See LICENSE (http://en.wikipedia.org/wiki/MIT_License) for details.
FAQs
Unknown package
We found that trinidad demonstrated a not healthy version release cadence and project activity because the last version was released a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
In this segment of the Risky Business podcast, Feross Aboukhadijeh and Patrick Gray discuss the challenges of tracking malware discovered in open source softare.
Research
Security News
A threat actor's playbook for exploiting the npm ecosystem was exposed on the dark web, detailing how to build a blockchain-powered botnet.
Security News
NVD’s backlog surpasses 20,000 CVEs as analysis slows and NIST announces new system updates to address ongoing delays.