Logjam is a Ruby on Rails application for finding performance hot spots and errors in web applications. Its primary target are Ruby on Rails applications, although it's possible to use it with other technologies.
It differs from other Rails monitoring solutions in that is is meant to be hosted in your own datacenter (or private cloud), for two reasons:
- it requires low latency between the monitored application and itself
- it stores extensive information for web requests, which might contain sensitive data
Logjam can produce various graphs and tables showing how much time and memory are consumed by different parts of your application. The interface is flexible enough to be able to provide answers to a great many questions you may have about your application, such as
- which actions in my application are most in need of optimization?
- in the worst case, how slow is my application?
- which actions are causing the most exceptions?
- what actions are causing the ruby heap to grow?
- which actions are making the most DB queries?
- which actions spend most of their time waiting for external services?
- how is my application behaving currently?
It also provides a live stream of performance and error data which is especially useful for supervising applications during deployments.
In addition, if you have several applications to monitor, and they call each other, Logjam can show you how often and when these calls occur.
Note that some of Logjam's features (e.g. those involving GC and memory data) depend on using a specially built version of Ruby (see https://github.com/skaes/rvm-patchsets).
A quick way of getting to understand what Logjam can do is to install and run a local development version, which monitors itself.
In the following, I assume you're a Rails developer.
- mongodb (see http://www.mongodb.org/downloads)
- zeromq (see http://zeromq.org/intro:get-the-software)
- memcached (see http://memcached.org/)
- a patched ruby for greater fun (see https://github.com/skaes/rvm-patchsets)
If you're on a Mac and are using MacPorts:
sudo port install mongodb memcached zmq libffi
I'm sure with Home Brew it's equally simple, but I haven't tried it.
In either case, you might need to create the mongodb data directory:
sudo mkdir -p /data/db
If you're using rvm, I recommend to install a patched ruby like so:
rvm install 2.1.2-railsexpress --patch railsexpress
rvm use 2.1.2-railsexpress@logjam --create
gem install bundler
Note: make sure to have a recent rvm version installed, otherwise it will not have up to date ruby patches. If in doubt, run
rvm get latest
git clone https://github.com/skaes/logjam_app.git
cd logjam_app
git submodule init
git submodule update
bundle
echo 2.1.2-railsexpress > .ruby-version
echo logjam > .ruby-gemset
mkdir -p log
Run
rake logjam:daemons:install
to create necessary service definitions under ./service.
Open a separate terminal session and run
foreman start
foreman will start all background processes which are necessary to enable self monitoring. If this gives you errors, e.g. because you have mongodb or memcached already running on your machine, comment out the corresponging lines in ./Procfile.
Then go back to the original session and run
rake logjam:db:update_known_databases
to initialize mongodb contents.
Run "rails s", then open the browser at http://localhost:3000/. You will be greeted with a "Dataset Empty" message. Reload the page to get some performance data. Then click around and explore the UI while your dataset continues to grow.
If you experience any problems following the steps above, please open a ticket.
Unsurprisingly, this requires you to configure Logjam to listen for request information and the application to send request data.
Logjam supports two different transport methods for sending request information from the application to Logjam:
- AMQP
- ZeroMQ
Both methods have their advantages and disadvantages:
-
AMQP requires a running message broker which acts as an intermediate, provides buffering and allows the logjam request importers to scale, but requires supervision.
-
With ZeroMQ, the application talks directly to the logjam request import daemon(s). Currently this doesn't scale as easily as the AMQP transport. This will be changed in the future.
For AMQP, I can recommend using RabbitMQ.
Application monitoring is configured through stream declarations in the file ./config/initializers/logjam_streams.rb. Such a declaration looks like this:
stream "app-env" do
importer do
type :zmq
port 9605
end
endHere "app" is the name which will be used by logjam internally and "env" the name of the environment (typically "production"). Note that each stream needs a separate port, on which the corresponding importer process will listen (interface 0.0.0.0).
For the AMQP transport:
stream "app-env" do
importer do
type :amqp
hosts "amqp-broker.at.your.org"
end
workers 3
endHere you can specify how many workers will be started by logjam to process request data.
- add the logjam_agent gem to the Gemfile of your application
- add an initializer for logjam_agent and specify transport method and endpoint
- add either the bunny gem or the ffi-rzmq gem to the Gemfile (based on which transport you chose)
- add an initializer for time_bandits and specify which metrics to track
Have a look at the corresponding files in logjam itself to get an idea on how to proceed.
Detailed explanation: coming soon ...
Coming soon ...
Stefan Kaes (skaes@railsexpress.de) and David Anderson (david@alpinegizmo.com).
Older versions of this code were MIT licensed. The current license version is GPLv3.
Copyright (c) 2009-2013 Stefan Kaes
Copyright (c) 2009 David Anderson
Copyright (c) 2009 XING AG
This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.