plone.recipe.cluster 1.0.0rc1

1.0.0rc1 (2010-03-18)

• Now, you can securely launch cluster from many (init.d script, fabric method, etc) : default pid file location is in buildout/var directory [tdesvenain]

• Now you can easily have many clusters in your buildout : pid filename depends on cluster part name [tdesvenain]

• make tests pass with latest zc.buildout versions (> 1.1.1) [tdesvenain]

0.1.0 (unreleased)

• Initial implementation of Linux daemon. [tarek]

• Created recipe with ZopeSkel. [tarek]

Supported options

The recipe supports the following options:

debug

default option is 0. When set to 1, allows tests to run on a non-blocking daemon.

start

Multiline option. Each line is a command line that will be called by the daemon when it is called with the start option. The commands must return immediatly, because the daemon waits for it before it launches the next command.

It is possible though, to push a command line in the background with the background: prefix.

stop

Multiline option. Each line is a command line that will be called by the daemon when it is called with the stop option. The commands must return immediatly, because the daemon waits for it before it launches the next command.

It is possible to ask the daemon to kill a given process, by providing a pid:value command line. Where value is either a PID number, either a text files that contains a pid.

The background: prefix is also available.

restart

Multiline option. Each line is a command line that will be called by the daemon when it is called with the restart option. The commands must return immediatly, because the daemon waits for it before it launches the next command.

It is possible to ask the daemon to kill a given process, by providing a pid:value command line. Where value is either a PID number, either a text files that contains a pid.

The background: prefix is also available.

pid-file

Defines the path to the PID file of the daemon.

Example usage

The cluster recipe allows you to create composite commands for your buildout. There are three commands:

• start

• stop

• restart

Each command is a variable with a list of commands to be run. The recipe then launches: - a daemon under Linux based system - a NT service under Windows

A typical usage for instance is to start zeo, zope and pound:

[buildout]

...

[cluster]
recipe = plone.recipe.cluster

poundctl = ${buildout:bin-directory}/pound -f ${buildout:directory}/parts/pound/etc/pound.cfg -c ${buildout:directory}/pound.pid


start =
    ${buildout:bin-directory}/zeoserver start
    ${buildout:bin-directory}/instance start
    ${cluster:poundctl}

stop =
    ${buildout:bin-directory}/zeoserver stop
    ${buildout:bin-directory}/instance stop
    pid:${buildout:directory}/pound.pid

restart =
    ${buildout:bin-directory}/zeoserver restart
    ${buildout:bin-directory}/instance restart
    ${cluster:poundctl}

Let’s try this:

>>> write('buildout.cfg',
... """
... [buildout]
... parts = mycluster
... index = http://pypi.python.org/simple
... [mycluster]
... recipe = plone.recipe.cluster
... debug = 1
... start =
...    background:${buildout:bin-directory}/script1
...    ${buildout:bin-directory}/script2 start
...
... stop =
...    ${buildout:bin-directory}/script2 stop
...
... restart =
...    background:${buildout:bin-directory}/script1
...    ${buildout:bin-directory}/script2 restart
...
... """)

Running the buildout gives us:

>>> print system(buildout)
Getting distribution for 'zc.recipe.egg'.
Got zc.recipe.egg ....
Installing mycluster.
Generated script '/.../bin/mycluster'.
<BLANKLINE>

Now let’s see the script that was created:

>>> script = join(sample_buildout, 'bin', 'mycluster')
>>> print open(script).read()
#!...python...
<BLANKLINE>
import sys
sys.path[0:0] = [
  ...
  ]
<BLANKLINE>
import plone.recipe.cluster.ctl
<BLANKLINE>
if __name__ == '__main__':
    plone.recipe.cluster.ctl.main(...)
<BLANKLINE>

Let’s create false scripts:

>>> script2 = join(sample_buildout, 'bin', 'script2')
>>> f = open(script2, 'w')
>>> f.write('echo script 2')
>>> f.close()
>>> import os
>>> os.chmod(script2, 0770)

>>> script1 = join(sample_buildout, 'bin', 'script1')
>>> f = open(script1, 'w')
>>> import sys
>>> f.write("""\
... #!%s
... import time
... #doesnt works with zc.buildout >= 1.1.1
... #while 1:
... #   time.sleep(0.1)
... """ % sys.executable)
>>> f.close()
>>> os.chmod(script1, 0770)

Let’s try to execute it:

>>> print system(script)
usage: /sample-buildout/bin/mycluster start|stop|restart|status
<BLANKLINE>

Let’s ask for the status:

>>> print system('%s status' % script)
Not running.

Ah right, let’s start it !::

>>> print system('%s start' % script)
Cluster is starting...
Running in background ...script1
Running ...script2 start
script 2
Child PIDs: ..., ...
Started with pid ...
Cluster is alive...
<BLANKLINE>

Let’s ask for the status:

>>> print system('%s status' % script)
Running.

We should not be able to start it twice::

>>> os.path.exists('%s/var/mycluster.pid' % sample_buildout)
True
>>> print system('%s start' % script)
<BLANKLINE>
Start aborded since pid file '...' exists.
<BLANKLINE>

Let’s stop it:

>>> print system('%s stop' % script)
Stopping PID ...
Stopping PID ...
Cluster is going down...
...

Let’s restart it:

>>> print system('%s restart' % script)
Could not stop, pid file '/sample-buildout/var/mycluster.pid' missing.
<BLANKLINE>

Oh right, it was stopped !