This repository holds legacy code related to The Marionette Collective project. That project has been deprecated by Puppet Inc and the code donated to the Choria Project.
Please review the Choria Project Website and specifically the MCollective Deprecation Notice for further information and details about the future of the MCollective project.
The shell agent allows you to start and manage shell commands via mcollective.
It allows the running of long-running processes with a mechanism to check in on the output from these long-running processes, which is independent of the mcollective daemon process (the daemon can be restarted without interrupting the processes)
To use this agent you need at least:
- MCollective 2.2.4
- Ruby 1.9 (for Process#spawn)
Please report any errors or make feature requests in the MCOP jira project
Please note: we do not recommend this agent as a way of building out your automation, for that you're still better off writing your own tailored agents that fit your use case. This agent is targeted at the ad-hoc needs that people occasionally have.
Follow the basic plugin install guide, taking all the code from lib and adding it to your MCollective $libdir
The agent should work without any additional configuration, though there are some options you can tune the mcollective server.cfg.
This is where the state used to track processes will live. By default this will be /var/run/mcollective-shell on Unix systems.
plugin.shell.state_directory = /opt/run/mcollective-shell
The mco shell application has several subcommands to start and manage
processes.
Runs a command and reports back. Use this for discrete short-living commands.
For long-running commands look at start or run --tail.
$ mco shell run dir
* [ ============================================================> ] 2 / 2
master:
bin dev home lib64 media opt root selinux srv tmp vagrant
boot etc lib lost+found mnt proc sbin src sys usr var
server2008r2a:
Volume in drive C has no label.
Volume Serial Number is DADF-75F9
Directory of C:\
09/22/2012 11:45 AM <SYMLINKD> manifests [\\vboxsrv\manifests]
09/22/2012 11:45 AM <SYMLINKD> modules [\\vboxsrv\modules]
07/13/2009 08:20 PM <DIR> PerfLogs
09/22/2012 11:42 AM <DIR> Program Files
03/27/2014 06:52 AM <DIR> Program Files (x86)
07/03/2014 07:42 AM <SYMLINKD> src [\\vboxsrv\C:_src]
03/27/2014 06:39 AM <DIR> Users
07/03/2014 07:42 AM <SYMLINKD> vagrant [\\vboxsrv\vagrant]
03/27/2014 06:41 AM <DIR> Windows
0 File(s) 0 bytes
9 Dir(s) 34,565,091,328 bytes free
Finished processing 2 / 2 hosts in 221.28 ms
Starts a command, shows you the output from it, kills the command when you interrupt with control-c, exits normally when the command exits.
$ mco shell -I /master/ run --tail vmstat 1
* [ ============================================================> ] 1 / 1
master stdout: procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu-----
master stdout: r b swpd free buff cache si so bi bo in cs us sy id wa st
master stdout: 0 1 445812 120584 5808 37348 34 29 52 48 39 47 6 1 93 0 0
master stdout: 1 0 445112 122692 5824 37332 2692 0 2692 84 911 2089 47 9 40 4 0
master stdout: 1 0 444848 122576 5824 37344 288 0 288 0 773 1914 48 5 47 0 0
master stdout: 0 0 444012 121320 5824 37348 1212 0 1212 0 823 1917 47 6 45 1 0
master stdout: 0 0 443984 121204 5824 37372 0 0 0 0 797 1796 52 5 43 0 0
master stdout: 0 0 438800 117244 5824 37360 3896 0 3896 0 910 2123 49 6 45 0 0
master stdout: 1 0 438768 117136 5840 37368 0 0 0 136 811 1926 48 6 45 0 0
^CAttempting to stopping cleanly, interrupt again to kill
Sending kill to master 6dad5cb9-57f7-46e0-bad7-07ab117369a5
Starts a command in the background and tells you the id that has been assigned
to it. You can then use mco shell watch, mco shell kill, mco shell list
to monitor this process and observe its output
$ mco shell -I /master/ start vmstat 1
* [ ============================================================> ] 1 / 1
master: 0dd67fac-734f-4824-8b4d-03100d4f9d07
Finished processing 1 / 1 hosts in 76.37 ms
Shows you the output of a command you previously started with mco shell start
$ mco shell watch 0dd67fac-734f-4824-8b4d-03100d4f9d07
* [ ============================================================> ] 2 / 2
master stdout: procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu-----
master stdout: r b swpd free buff cache si so bi bo in cs us sy id wa st
master stdout: 2 0 431448 110704 8484 40644 34 29 52 48 40 47 6 1 93 0 0
Show a list of running jobs.
$ mco shell list -v
* [ ============================================================> ] 2 / 2
master:
0dd67fac-734f-4824-8b4d-03100d4f9d07
1fd3961a-f48d-4119-b988-146b490a5ca3
d174e20b-9cdb-4c14-9f34-fd29995f30cb
ea809b20-3123-46b4-bf59-10ff7251ca9b
Finished processing 2 / 2 hosts in 142.34 ms
Kill a running job.
$ mco shell kill 0dd67fac-734f-4824-8b4d-03100d4f9d07
* [ ============================================================> ] 2 / 2
Finished processing 2 / 2 hosts in 170.17 ms