mco doesn't deploy modules when using forge functionality of Puppetfile about puppet-r10k HOT 13 CLOSED

I think there may be an issue with Puppetfile's that contain forge entries instead of git entries can you confirm which on you are using?

from puppet-r10k.

both, actually.

forge 'http://forge.puppetlabs.com'

mod 'puppetlabs/apache'
mod 'puppetlabs/mysql'
mod 'puppetlabs/vcsrepo'
mod 'puppetlabs/stdlib'
mod 'hunner/wordpress'
mod 'nanliu/staging'
mod 'seteam/opencsw'

mod 'xinetd',
:git => 'https://github.com/vchepkov/puppetlabs-xinetd.git',
:ref => 'development'

mod 'ntp',
:git => 'https://github.com/vchepkov/puppetlabs-ntp.git',
:ref => 'development'

from puppet-r10k.

@vchepkov are you still having this issue?

from puppet-r10k.

Yes, I even changed r10k version to 1.3.2, default was 1.1.0
But it didn't help. To test it I remove a module from an environment :

# rm -rf /etc/puppetlabs/puppet/environments/production/modules/staging/
# sudo -u peadmin /opt/puppet/bin/mco r10k synchronize
...
Finished processing 1 / 1 hosts in 2622.32 ms

And I see log entry in mcollective-audit.log on master, but yet no module

# tail /var/log/pe-mcollective/mcollective-audit.log
... agent=r10k action=synchronize data={:process_results=>true}

# ls /etc/puppetlabs/puppet/environments/production/modules/staging/
ls: cannot access /etc/puppetlabs/puppet/environments/production/modules/staging/: No such file or directory

I enabled prerun_command, but this brought up some other unexpected consequences. We are still working on the support ticket #6470, but it looks like inclusion of r10k into prerun_command breaks PE upgrade process.

from puppet-r10k.

Are you seeing the same behaviour with ntp, I am wondering if this is an issue with the Puppet Module Tool running via mcollective.

from puppet-r10k.

got modules are installed, i.e. NTP is

from puppet-r10k.

Stupid speller - git modules

from puppet-r10k.

I think the large majority of people use the git functionality who also use the webhook, so I need to try and trace this issue down, can you provide me with your version of PE?

from puppet-r10k.

Also any other relevant info such as OS version, selinux status. I think this has to do with the puppet module tool running under mcollective. So this for instance could be impacted by things like a proxy

from puppet-r10k.

I am using PE 3.3.2. RedHat 6.5, SELinux is disabled, no proxies.

from puppet-r10k.

After doing a bit of debugging (no strace or anything), I believe this may have something to do with r10k calling "puppet module" and assuming puppet is in the PATH. This is just a guess, but I've noticed the same issue happening when calling r10k from cron with a limited sh shell. As soon as we modified the cron job to have puppet in it's PATH, everything worked as expected.

After creating a link to the puppet binary to /usr/bin/puppet, everything worked as expected. This isn't a full blown formal proof or anything, but it's a good sign that the PATH is the issue.

from puppet-r10k.

acidprime@7715485

from puppet-r10k.

Released in https://github.com/acidprime/r10k/releases/tag/v2.5.3

from puppet-r10k.