Reports from puppet

[Maxim Nikolaev]

Hi

I'm using Puppet with Dashboard and PuppetDB and Puppetdb board. I can see all nodes and rfeports.
Problem is that when puppet fail to run on instance - i get report "unchanged" instead of "fail".

For ex. I've changed postfix manifest to install package postfix1. Puppet failed to run:

Error: Could not retrieve catalog from remote server: Error 400 on SERVER: Invalid relationship: File[/etc/postfix/main.cf] { require => Package[postfix] }, because Package[postfix] doesn't seem to be in the catalog
But I got unchanged report instead of failed also in dashboard and in puppetdb.

Puppet: 3.6.2
Facter: 2.1.0
OS: Amazon Linux

[Ken Barber]

Hi Maxim,

This is not directly reproducible by myself today:
https://gist.github.com/kbarber/c6941099bea07096361e ...

Perhaps something in your puppet.conf is doing this, I could imagine
something like:

usecacheonfailure = true

Causing this to happen, but I can't reproduce the exact same conditions myself.

Here is my basic puppet.conf fwiw:

# cat /etc/puppet/puppet.conf
[main]
logdir=/var/log/puppet
vardir=/var/lib/puppet
ssldir=/var/lib/puppet/ssl
rundir=/var/run/puppet
factpath=$vardir/lib/facter

[agent]
report = true

[master]
ssl_client_header = SSL_CLIENT_S_DN
ssl_client_verify_header = SSL_CLIENT_VERIFY
storeconfigs = true
storeconfigs_backend = puppetdb
reports = store,puppetdb
trusted_node_data = true

Perhaps you can share your so we can see what settings may be causing it.

ken.

> --
> You received this message because you are subscribed to the Google Groups
> "Puppet Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to puppet-users...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/puppet-users/29057799-8d81-4272-b02d-8df2f8ba38f7%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Maxim Nikolaev]

No. I haven't this feature.
my puppet.conf looks like this:

[main]
modules = /etc/puppet/modules
hiera_config = /etc/puppet/hiera.yaml
pluginsync    = true
server = HOSTNAME
#port = 8081

[agent]
server = HOSTNAME
report = true
classfile = $vardir/classes.txt
localconfig = $vardir/localconfig

[master]
certname = HOSTNAME
dns_alt_names = HOSTNAME
autosign = true
reports = store, http, puppetdb, tagmail
reporturl = http://HOSTNAME/reports/upload
hiera_config = /etc/puppet/hiera.yaml

storeconfigs = true
storeconfigs_backend = puppetdb

I found discussion about this bug:

http://projects.theforeman.org/issues/3851

But I want to know if there is any workaround.

If catalog fails on client side  - I can see error reports.

[José Luis Ledesma]

Which version of puppedb are you running? If I'm not wrong, from puppetdb 2.1 catalog compilation errors are stored , but not for previous versions.

Regards,

To view this discussion on the web visit https://groups.google.com/d/msgid/puppet-users/f9cde2d1-0d85-49ec-a5bd-2582dd01ff16%40googlegroups.com.

[Ken Barber]

Are we even sure this is a PuppetDB problem? And not an issue with the
Dashboard itself? As I've shown, the report gets stored in PuppetDB as
a failure. Having said that, if there is an issue in the summary of
the report we don't store those stats ourselves in PuppetDB.

Maxim, I'd do the test I provided in my gist, that is - the curl
request against PuppetDB to confirm this is true for PuppetDB at
least. For 2.1.0 and above this should work fine. At the very least we
need to determine where this fault is. It sounds like PUP-283, but the
latest version of Puppet Board shouldn't suffer this issue since its
not querying against the Dashboard, but PuppetDB directly.

ken.

> https://groups.google.com/d/msgid/puppet-users/CAF_B3dcBY5T5sEisZrK%3DgyF9woaZiCqdT_hw5zFxDCCUyAr1yQ%40mail.gmail.com.

[Spencer Krum]

Also, what version of puppetboard is being used?

To view this discussion on the web visit https://groups.google.com/d/msgid/puppet-users/CAE4bNTnU6YffqzkZxxk%3DX8AAW8_z7XiUHORc%3DWnu8N4T4eJNRw%40mail.gmail.com.

[Maxim Nikolaev]

HI

I'm using puppetdb-2.1.0-1.el6.noarch
Puppetboard installed from pip yesterday, so I suppose it's also last.

I get mail about errors, so I know that report is generated. But I can't see it not in Puppetboard (
Overvie). When I check report in Node tab - I see that it's empty.

Error on client:

 puppet agent --no-daemonize --verbose --onetime
 Info: Retrieving pluginfacts
 Info: Retrieving plugin
 Info: Loading facts in /var/lib/puppet/lib/facter/puppet_vardir.rb
 Info: Loading facts in /var/lib/puppet/lib/facter/root_home.rb
 Info: Loading facts in /var/lib/puppet/lib/facter/facter_dot_d.rb
 Info: Loading facts in /var/lib/puppet/lib/facter/pe_version.rb
 Error: Could not retrieve catalog from remote server: Error 400 on SERVER: Could not find class rabbitmq::rabbitmq for mcollective-useast-00-d6f9.ec2.internal on node mcollective-useast-00-d6f9.ec2.internal
 Notice: Using cached catalog
 Error: Could not retrieve catalog; skipping run

In Dashboard I can see error, but in Puppetboard - not. Puppetboard sign node as unchanged and remove it from Overview tab. I can see it in Node, but again as unchanged and not failed.

[Felix Frank]

On 10/14/2014 10:23 AM, Georg Brunmayr wrote:
> Hi,
>
> sorry for the necro but could somebody finally resolve this issue?
> I currenty have the same problem in an EC2 environment - i get mails for
> failures but the puppetboard does not show the node as failed.
>
> Thx
> Georg

I can imagine that puppetboard would need to incorporate a fix similar
to that for http://projects.theforeman.org/issues/3851

Other than that, reporting seems to be somewhat problematic in the case
of failed compilations.