forked from kim/evented-gearman-ruby
-
Notifications
You must be signed in to change notification settings - Fork 1
/
HOWTO
146 lines (106 loc) · 6.27 KB
/
HOWTO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
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
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
= GEARMAN
"Gearman provides a generic application framework to farm out work to other
machines or processes that are better suited to do the work. It allows you to
do work in parallel, to load balance processing, and to call functions between
languages. It can be used in a variety of applications, from high-availability
web sites to the transport of database replication events. In other words, it
is the nervous system for how distributed processing communicates."
- http://www.gearman.org/
== Setting up a basic environment
A very basic German environment will look like this:
----------
| Client |
----------
|
--------------
| Job Server |
--------------
|
----------------------------------------------
| | | |
---------- ---------- ---------- ----------
| Worker | | Worker | | Worker | | Worker |
---------- ---------- ---------- ----------
And the behavior will be the following:
* JobServer: Acts as a message passing point.
* Client: Sends tasks to the JobServer. Will be connected to only one JobServer
in case more than one exits for failover purposes.
* Worker: Anounce his 'abilities' to the JobServer and waits for tasks.
For the JobServer we recommend to use the offical Perl version, there's also a
more performant C implementation of the server with support for persistent
queues, bells and whistles but is not stable enough for production use at the
time of this document was wrote.
The Client and the Worker can be implemented in any language. This way you can
send tasks from a Ruby client server, to a Perl or C worker in order to get
better performance.
== Installing the required software
For the JobServer we recommend to use the offical Perl version, to install it:
* Mac OS X: sudo port install p5-gearman-server
* Debian/Ubuntu: sudo apt-get install gearman-server
To get the Ruby libraries by Xing:
git clone git://github.com/xing/gearman-ruby.git
== Gearman demo
Now you're ready for you first experience with Gearman. In the cloned repository
you'll find an 'examples' directory.
Run the 'gearman_environment.sh' to build an environment like the one showed in
the diagram above.
* Client: Will ask you for an arithmetic operation, like: 2+3
The code of the client is in: 'examples/calculus_client.rb'
* JobServer: The Perl server.
* Workers: You'll have 4 worker, one for each of the basic arithmetic
operations.
The code of the worker is in: 'examples/calculus_worker.rb'
There are other demos in the examples folder you can give a look at. Each demo usually
consist in the client and server scripts.
=== Creating clients and tasks
In order to get a job scheduled by a Gearman server using the gearman ruby library, there
are three main objects you must interact with: Gearman::Client, Gearman::Task and Gearman::TaskSet.
Let's review all of them briefly:
- Gearman::Client -> the portion of the library storing the data about the connection to
the Gearman server.
- Gearman::Task -> a job execution request that will be dispatched by the Gearman server to
worker and whose result data will be returned to the client.
- Gearman::TaskSet -> a collection of tasks to be executed. The Taskset object will track the
execution of the tasks with the info returned from the Gearman server
and notify the client with the results or errors when all the tasks
have completed their execution.
To send a new task to the Gearman server, the client must build a new Gearman::Task object, add it to
a Gearman::TaskSet that must hold a reference to a Gearman::Client and send the wait message to
the TaskSet.
The following code taken from examples/client.rb shows the process:
----------------------------------------------------
servers = ['localhost:4730', 'localhost:4731']
client = Gearman::Client.new(servers)
taskset = Gearman::TaskSet.new(client)
task = Gearman::Task.new('sleep', 20)
task.on_complete {|d| puts d }
taskset.add_task(task)
taskset.wait(100)
----------------------------------------------------
The name of the function to be executed is the first parameter to the constructor of the Task object.
Take into account that the string you pass as a parameter will be used 'as it' by the Gearman server
to locate a suitable worker for that function.
The second parameter is the argument to be sent to the worker that will execute, if the arguments for
the task are complex, a serialization format like YAML or XML must be agreeded with the workers.
The last and optional parameter is a hash of options. The following options are currently available:
- :priority -> (:high | :low) the priority of the job, a high priority job is executed before a low on
- :background -> (true | false) a background task will return no further information to the client.
The execution of a task in a Gearman remote worker can fail, the worker can throw an exception, etc.
All these events can be handled by the client of the ruby library registering callback blocks.
The following events are currently available:
- on_complete -> the task was executed succesfully
- on_fail -> the task fail for some unknown reason
- on_retry -> after failure, the task is gonna be retried, the number of retries is passed
- on_exception -> the remote worker send an exception notification, the exception text is passed
- on_stauts -> a status update is sent by the remote worker
In order to receive exception notifications in the client, this option must be sent in the server, the
method option_request can be used this task. The following example, extracted from the examples/client_exception.rb
demo script shows the process:
----------------------------------------------------
client = Gearman::Client.new(servers)
#try this out
client.option_request("exceptions")
----------------------------------------------------
This feature will only works if the server and workers have implemented support for the OPT_REQ and WORK_EXCEPTION
messages of the Gearman protocol.
Enjoy.