On Monday, June 18, 2012 4:53:23 AM UTC-5, Javi wrote:
>
> Scenario: install SNMP (agent and daemon), change config file and restart
> of service.
>
> What's wrong? The logs showed that AFTER the new config file is applied,
> the service (snmpd) is restarted, but looks like this is not true.
>
The log you presented shows that Puppet took the action it thought
appropriate for refreshing the service, and that action returned a success
code. If that didn't actually (re)start the service then probably your
declarations for it are inconsistent with the actual service implementation
(most likely with the initscript), or else something else happening on the
node is interfering.
Jun 15 17:08:21 test-3 puppet-agent[1219]:
> (/Stage[main]/Apt::Service/Exec[/usr/bin/apt-get update]/returns) executed
> successfully
> Jun 15 17:08:44 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Install/Package[snmp]/ensure) ensure changed 'purged'
> to 'present'
>
[... many complaints from snmpd ...]
Why is snmpd logging anything at that point? Does installing the package
automatically start the service? That would be horrible!
Jun 15 17:09:01 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Install/Package[snmpd]/ensure) ensure changed 'purged'
> to 'present'
> Jun 15 17:09:01 test-3 puppet-agent[1219]: FileBucket adding
> /etc/snmp/snmpd.conf as {md5}a5007383dd9c4ef73500e3df8c080665
> Jun 15 17:09:01 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Config/File[/etc/snmp/snmpd.conf]) Filebucketed
> /etc/snmp/snmpd.conf to puppet with sum a5007383dd9c4ef73500e3df8c080665
> Jun 15 17:09:02 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Config/File[/etc/snmp/snmpd.conf]/content) content
> changed '{md5}a5007383dd9c4ef73500e3df8c080665' to
> '{md5}6a797811e82b5f411af1093ea6336a04'
> Jun 15 17:09:02 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Config/File[/etc/snmp/snmpd.conf]) Scheduling refresh
> of Service[snmpd]
> Jun 15 17:09:02 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Service/Service[snmpd]/ensure) ensure changed 'stopped'
> to 'running'
>
The "refresh" Puppet performed was an attempt to change the service from
the "stopped" state to the "running" state. For Debian-family distros,
that probably means running '/etc/init.d/snmpd start'. Note: not *re*start,
because Puppet doesn't think the service is running yet. That's
suspicious, because snmpd was just observed logging all sorts of complaints
about its config file.
> Jun 15 17:09:02 test-3 puppet-agent[1219]:
> (/Stage[main]/Snmp::Service/Service[snmpd]) Triggered 'refresh' from 1
> events
> Jun 15 17:09:02 test-3 puppet-agent[1219]: Finished catalog run in 73.61
> seconds
>
So the order of operations appears correct, but the refresh of
Service['snmpd'] doesn't seem to take. I see a couple of plausible
explanations:
- The snmpd service is already running when Puppet tries to refresh it,
but Puppet thinks it's stopped. Puppet therefore issues the 'start'
command to the initscript, which typically does nothing if the service is
already running. I'd put my bet here.
- The initscript 'start' command returns an exit code indicating
success, even though it failed to start the service for some unknown
reason.
Let's focus now on the service declaration:
puppet-1.dev.jj.com:/etc/puppet/modules/snmp/manifests/service.pp
>
> class snmp::service {
> service { $snmp::params::snmp_service_name:
> ensure => running,
> hasstatus => true,
> hasrestart => true,
>
Those 'hasstatus' and 'hasrestart' declarations tell Puppet that the
service's init script supports the 'restart' and 'status' commands, and
that they return codes at least approximately consistent with those
specified by the LSB. In particular, Puppet pays attention to whether the
those commands' exit codes are zero or nonzero. You should verify that the
'status' command returns 0 if and only if the service is already running,
and that the 'restart' command returns 0 if and only if the service is
successfully (re)started.
> # Boot time
> enable => true,
> require => Class["snmp::config"],
>
That require is mostly redundant, but harmless.
> }
> }
>
Overall, the manifests look good, subject to the caveats I already
discussed.
If the problem is that installing the package starts snmpd behind Puppet's
back, then probably the situation will resolve itself on the second Puppet
run. Alternatively, perhaps there is a way to prevent package install from
starting the service.
If the problem is that the init script's 'status' or 'restart' command is
not working properly, then you can either fix the script between installing
the package and managing the service, or you can instead specify an
alternative start and/or status command via the Service's 'status' and
'restart' parameters.
John
--
You received this message because you are subscribed to the Google Groups
"Puppet Users" group.
To view this discussion on the web visit
https://groups.google.com/d/msg/puppet-users/-/oN9V3JEezzAJ.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en.
