README.pod

August 31, 2012 · View on GitHub

=head1 NAME

Pogo - Run commands on many hosts in a controlled manner

=head1 DESCRIPTION

Pogo is a highly scalable system for running arbitrary commands on many hosts in a controlled manner.

It is mostly used for quick mass software deployments on server farms while making sure only an allowed number of nodes are upgraded in parallel to ensure business continuity.

=head2 Project Setup

Pogo is hosted on Github at

https://github.com/ytoolshed/pogo

The latest stable version can be found on the master branch. The project is automatically being tested on every commit, using travis-ci's service:

http://travis-ci.org/#!/ytoolshed/pogo

=head2 Architecture

Pogo consists of several components, which can be all running on the same system, or, in order to scale it, be replicated and even be installed on many distributed hosts. Those components are

=over 4

=item Client

Users submit jobs to pogo using the client, which in turn contacts the API.

=item API

Takes requests via HTTP from the client and forwards them to a dispatcher.

=item Dispatcher

Takes job requests from the API, figures out constraints, and determines single tasks the job consists of. It then assigns tasks to workers, watches their individual completion and keeps track of overall job completion. Dispatchers can be queried by the API to determine the status of a given job.

=item Worker

Takes a task (like "ssh to a host and run this command") from the dispatcher, executes it and reports back the result. Can handle many tasks concurrently.

=back

=head2 Security

To make sure dispatchers and workers communicate over secure channels, and enable them to authenticate each other (is a connecting worker really an authorized worker, or is the dispatcher it's connecting to really an authorized dispatcher?), Pogo uses SSL server and client certs. See LPogo::Security for details.

=head1 NAME

pogo-one - All-In-One Pogo

=head1 SYNOPSIS

$ cat t/cfgs/one.cfg
tag:
  # no tags
sequence:
  - host3
  - host2
  - host1

  # specify two hosts 1 and 2, but config says to run them in 
  # reverse order
$ perl -Ilib bin/pogo-one --password --targets host1,host2 \
    --config t/cfgs/one.cfg --run-command 'date'
Password: ****
host2: Running target
host2: Success
host1: Running target
host1: Success

=head1 DESCRIPTION

C is a simple utility that loads all the usually distributed Pogo components into one single process and runs it from the command line. In this way, you can test Pogo functions like constraints and sequences without having to set up the different components of an entire production system. It is mainly used for testing, but also helps with small deployment jobs.

=head2 PASSWORDS

By default, C assumes that you have access to the target boxes without having to type your password or your ssh keyring's passphrase. To have C collect the password and use it to sign in on the target, use the C<--password> options

$ pogo-one --password ...
Password: ****
...

=head2 OPTIONS

=over 4

=item C<--config file>

Location of configuration file that defineds Pogo tags, sequences, and constraints.

=item C<--password>

If specified, c will collect the user's target password at a command line prompt, and use it to log into the target machines.

=item C<--targets>

All hosts to run the command on. Order and parallelism is determined by the configuration file.

=item C<--run-command>

The command to run on all hosts.

=back

=head1 NAME

pogo-schedule - Print out the schedule for a job

=head1 SYNOPSIS

$ cat t/cfgs/triple.cfg
# sample configuration file

tag:
  colo:
    one:
      - host1
      - host2
      - host3
    two:
      - host4
      - host5
      - host6
    three:
      - host7
      - host8
      - host9
sequence:
  foo:
    - $colo.one
    - $colo.two
  bar:
    - $colo.three

$ perl -Ilib bin/pogo-schedule t/cfgs/triple.cfg
.--------------------------------------------------------------.
|                           Schedule                           |
+------------------+---------------------+---------------------+
| Thread           | slot-1              | slot-2              |
+------------------+---------------------+---------------------+
| thread-000000000 | [foo.$colo.one]     | [foo.$colo.two]     |
|                  | host3, host2, host1 | host6, host5, host4 |
| thread-000000001 | [bar.$colo.three]   |                     |
|                  | host9, host8, host7 |                     |
| unconstrained    | [unconstrained]     |                     |
'------------------+---------------------+---------------------'

=head1 DESCRIPTION

This utility pretty-prints a schedule for a job, as it would be executed based on a Pogo configuration file. It uses the actual Pogo scheduler in Pogo::Scheduler::Classic and asks it for an execution plan just as if we submitted a Pogo job for all hosts present in the configuration.

=head1 LICENSE

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and imitations under the License.

=head1 AUTHORS

Mike Schilli m@perlmeister.com Ian Bettinger ibettinger@yahoo.com

Many thanks to the following folks for implementing the original version of Pogo:

Andrew Sloane andy@a1k0n.net, Michael Fischer michael+pogo@dynamine.net, Nicholas Harteau nrh@hep.cat, Nick Purvis nep@noisetu.be, Robert Phan robert.phan@gmail.com, Srini Singanallur ssingan@yahoo.com, Yogesh Natarajan yogesh_ny@yahoo.co.in