cmd.run with watch or onchanges does not work or I don't understand

Tom K.

unread,

Nov 17, 2016, 9:55:26 AM11/17/16

to Salt-users

I've some misunderstanding with cmd.run and watch/onchanges

I've setup a nrpe nagios server with salt. But I just want do reload/restart the server if there are changes to some configs.

My testings are with -watch and - onchanges. But everytime the state runs, the restartIfNeeded is executed.

cmd.wait is working but it's deprecated, so I don't want to use it anymore.

Thanks,

Tom

salt 2016.3.1 (Boron)

/etc/nagios/nrpe.cfg:

file.managed:

- source: salt://default-files/nrpe.cfg.jinja

- user: root

- group: root

- mode: 644

- template: jinja

- context:

collector: {{ possibleCollector }}

/etc/nagios/nrpe.d/base.cfg:

file.managed:

- source: salt://default-files/nrpe-base.cfg.jinja

- user: root

- group: root

- mode: 644

- template: jinja

- context:

w1: {{ grains['num_cpus'] * 1.5 }}

w5: {{ grains['num_cpus'] * 1.5 }}

w15: {{ grains['num_cpus'] * 1.5 }}

c1: {{ grains['num_cpus'] * 3 }}

c5: {{ grains['num_cpus'] * 2 }}

c15: {{ grains['num_cpus'] * 2 }}

restartIfNeeded:

cmd.run:

- name: 'service nagios-nrpe-server restart'

- user: root

- group: root

- watch:

- file: /etc/nagios/nrpe.d/base.cfg

- file: /etc/nagios/nrpe.cfg

Results in

networkingtest-001.off.loc:

----------

ID: nrpe

Function: pkg.installed

Name: nagios-nrpe-server

Result: True

Comment: Package nagios-nrpe-server is already installed

Started:

Duration:

Changes:

----------

ID: nrpe

Function: service.running

Name: nagios-nrpe-server

Result: True

Comment: Service nagios-nrpe-server has been enabled, and is in the desired state

Started:

Duration:

Changes:

----------

nagios-nrpe-server:

True

----------

ID: /etc/nagios/nrpe.cfg

Function: file.managed

Result: True

Comment: File /etc/nagios/nrpe.cfg is in the correct state

Started:

Duration:

Changes:

----------

ID: /etc/nagios/nrpe.d/base.cfg

Function: file.managed

Result: True

Comment: File /etc/nagios/nrpe.d/base.cfg is in the correct state

Started:

Duration:

Changes:

----------

ID: restartIfNeeded

Function: cmd.run

Name: service nagios-nrpe-server restart && echo "watch"

Result: True

Comment: Command "service nagios-nrpe-server restart" run

Started:

Duration:

Changes:

----------

pid:

22303

retcode:

0

stderr:

stdout:

Summary for networkingtest-001.off.loc

------------

Succeeded: 5 (changed=2)

Failed: 0

------------

Thomas Phipps

unread,

Nov 17, 2016, 12:01:31 PM11/17/16

to Salt-users

see this https://github.com/saltstack/salt/issues/29492#issuecomment-162967147

also since you are just calling service restart any reason service.running with a watch is not working for you?

--
You received this message because you are subscribed to the Google Groups "Salt-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to salt-users+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Thomas Phipps

Linux/Openstack/Logstash

(702) 518-0131

Tom K.

unread,

Nov 17, 2016, 12:20:48 PM11/17/16

to Salt-users

Hi Thomas,

you are right. For the restart if that service it works fine with service.running.

But what about something like

jetty_restart:

cmd.run:

- name: '/usr/local/share/jetty9/bin/jetty.sh stop ; sleep 2 ; /usr/local/share/jetty9/bin/jetty.sh start'

- user: root

- group: root

- onchanges:

- file: /usr/local/share/jetty9/bin/jetty.sh

This should restart if some changes in jetty.sh are made. For example vm settings.

Thomas Phipps

unread,

Nov 17, 2016, 12:28:50 PM11/17/16

to Salt-users

yes, as long as you have a file.managed that handles /usr/local/share/jetty9/bin/jetty.sh yes

Tom K.

unread,

Nov 18, 2016, 5:12:14 AM11/18/16

to Salt-users

Hm,

Hi,

what about that setup? It doesn't matter of I do a -onchanges or -watch by "apply_sysctl"

State "apply_sysctl" always executed.

/etc/sysctl.conf:

file.managed:

- name: /etc/sysctl.conf

- source: salt://default-files/sysctl.conf

- user: root

- group: root

apply_sysctl:

cmd.run:

- name: 'sysctl -p'

- user: root

- group: root

- watch:

- file: /etc/sysctl.conf

----------

ID: /etc/sysctl.conf

Function: file.managed

Result: True

Comment: File /etc/sysctl.conf is in the correct state

Started:

Duration:

Changes:

----------

ID: apply_sysctl

Function: cmd.run

Name: sysctl -p

Result: True

Comment: Command "sysctl -p" run

Started:

Duration:

Changes:

----------

pid:

24974

retcode:

0

stderr:

stdout:

----------

Thanks,

Tom

Dmitry Golubenko

unread,

Nov 18, 2016, 5:30:43 AM11/18/16

to salt-...@googlegroups.com

В Птн, 18/11/2016 в 02:12 -0800, 'Tom K.' via Salt-users пишет:

> Hm,
> Hi,
>
>
> what about that setup? It doesn't matter of I do a -onchanges or
> -watch by "apply_sysctl"
> State "apply_sysctl" always executed.

it's because cmd.run always executed, it can't `watch'. try cmd.wait

Tom K.

unread,

Nov 18, 2016, 5:35:33 AM11/18/16

to Salt-users, dmitry.g...@xiag.ch

Hi Dmitry,

cmd.wait is working but it's deprecated, so I don't want to use it anymore.

Thomas Phipps

unread,

Nov 21, 2016, 10:36:21 AM11/21/16

to salt-...@googlegroups.com

please do NOT use watch with cmd.run ever. watch still runs the sls. always use onchanges with cmd.run if you have to use cmd.run and want it to have a watch like functionality.

also post the output of onchanges. you are still only showing watch. which will never do what you want.

the main difference between watch and onchanges is watch runs the sls depending on

this is the result i get when i use onchanges

[root@salt01 testing]# salt 'salt01*' state.apply testing.test

salt01.wolfcloud.bad4.us:

----------

ID: test_file

Function: file.managed

Name: /tmp/file

Result: True

Comment: File /tmp/file is in the correct state

Started: 09:06:39.964343

Duration: 19.599 ms

Changes:

----------

ID: echo_result

Function: cmd.run

Name: echo test

Result: True

Comment: State was not run because none of the onchanges reqs changed

Started:

Duration:

Changes:

here is my sls

test_file:

file.managed:

- name: /tmp/file

- source: salt://testing/files/test

- user: root

- group: root

echo_result:

cmd.run:

- name: 'echo test'

- user: root

- onchanges:

- file: test_file

Tom K.

unread,

Nov 23, 2016, 11:44:13 AM11/23/16

to Salt-users

Hi Thomas,

hm, what am I doing wrong?? I Think I use onchanges, but it fires up the cmd.run to restart my firewall

/etc/firewall.sh:

file.managed:

- name: /etc/firewall.sh

- source: salt://default-files/firewall/firewall.sh

- user: root

- group: root

- mode: 755

start_firewall:

cmd.run:

- name: '/etc/init.d/firewall restart'

- user: root

- group: root

- onchanges:

- file: /etc/firewall.sh

----------

ID: /etc/firewall.sh

Function: file.managed

Result: True

Comment: File /etc/firewall.sh is in the correct state

Started:

Duration:

Changes:

----------

ID: start_firewall

Function: cmd.run

Name: /etc/init.d/firewall restart

Result: True

Comment: Command "/etc/init.d/firewall restart" run

Started:

Duration:

Changes:

----------

pid:

15954

retcode:

0

stderr:

stdout:

-e Stopping Firewall ...

-e Starting Firewall ...

Thomas Phipps

unread,

Nov 23, 2016, 12:33:10 PM11/23/16

to Salt-users

at this point i am not sure. maybe you are running into a bug that was on 2016.3.1. however i couldn't find a bug on that version. i copied your state into my setup and it worked without issue.

there is one thing interesting. your file.managed state does not have a start time or a duration time. while mine does even on the unchanged one. one thing you can try doing is using -l debug and salt-call to see more about what salt is trying to do.

Reply all

Reply to author

Forward