unable to add pulp_rpmrepo (undefined method `each' for nil:NilClass) #245

TJM · 2017-05-24T01:17:01Z

We created our pulp server today and tried to use the pulp_rpmrepo provider to configure it, but it is getting the following error: (end of puppet agent -t --debug)

Debug: Prefetching api resources for pulp_rpmrepo
Debug: Storing state
Debug: Stored state in 0.05 seconds
Error: Failed to apply catalog: undefined method `each' for nil:NilClass
Debug: Dynamically-bound server lookup failed, falling back to report_server setting
Debug: Dynamically-bound port lookup failed; falling back to report_port setting
Debug: Using cached connection for https://mbi-foreman-01.utmb.lan:8140
Debug: Caching connection for https://mbi-foreman-01.utmb.lan:8140
Debug: Closing connection for https://mbi-foreman-01.utmb.lan:8140

Here is the relevant data:

---
pulp::debugging_mode: true
pulp::log_level: DEBUG
pulp::ssl_verify_client: optional
pulp::https_cert: /etc/puppetlabs/puppet/ssl/certs/%{::fqdn}.pem
pulp::https_key: /etc/puppetlabs/puppet/ssl/private_keys/%{::fqdn}.pem
pulp::https_chain: /etc/puppetlabs/puppet/ssl/certs/ca.pem
pulp::admin::verify_ssl: false
pulp::admin::logging_filename: /var/log/pulp/admin.log
pulp::admin::logging_call_log_filename: /var/log/pulp/server_calls.log

profile::pulp::rpmrepos:
  centos-7-x86_64-base:
    display_name: CentOS 7 x86_64 Base
    feed: http://mirror.centos.org/centos/7/os/x86_64/
    relative_url: centos/7/os/x86_64/
    remove_missing: true
    retain_old_count: 2
    serve_http: true
    serve_https: true
    validate: true
  centos-7-x86_64-updates:
    display_name: CentOS 7 x86_64 Updates
    feed: http://mirror.centos.org/centos/7/updates/x86_64/
    relative_url: centos/7/updates/x86_64/
    remove_missing: true
    retain_old_count: 2
    serve_http: true
    serve_https: true
    validate: true
  centos-7-x86_64-epel:
    display_name: CentOS 7 x86_64 EPEL
    feed: http://download.fedoraproject.org/pub/epel/7/x86_64/
    relative_url: epel/7/x86_64/
    remove_missing: true
    retain_old_count: 2
    serve_http: true
    serve_https: true
    validate: true

Here is the profile/manifests/pulp.pp:

class profile::pulp (
  $rpmrepos = {},
) {

# Repo: You will have to use upstream until pulp is online, then it can point to itself
#  include ::repos::pulp2
  include ::pulp::repo::upstream

  include ::apache
  include ::apache::mod::proxy
  include ::apache::mod::headers

  include ::pulp
  include ::pulp::admin

  # RPM Repos
  create_resources('pulp_rpmrepo', $rpmrepos)

  firewall { '100 INPUT allow http(s) from all':
    dport  => [ '80', '443' ],
    proto  => 'tcp',
    action => 'accept',
    chain  => 'INPUT',
  }

  firewall { '100 INPUT allow http(s) from all IPv6':
    dport    => [ '80', '443' ],
    proto    => 'tcp',
    action   => 'accept',
    chain    => 'INPUT',
    provider => 'ip6tables',
  }
}

In addition, I had to SSH in and run pulp-admin login -u admin using the "default_password" to populate /root/.pulp/user-cert.pem. Should that be necessary? (or am I missing something, the README.md is a bit thin) :)

~tommy

The text was updated successfully, but these errors were encountered:

ekohl · 2017-05-24T10:41:22Z

Do you happen to have a line number where it went wrong?

Other than that I'm surprised you have to include apache and those modules since we include what we need. If we don't, that's a bug.

For the README expansion see #242.

TJM · 2017-05-25T03:14:01Z

@ekohl - Admittedly that part of the profile was 'inherited' config. It is probably not necessary. I will try removing it. It doesn't really give a line number with the error, but it works fine if I change the yaml key (such that profile::pulp::rpmrepos becomes undefined)

I even tried shortening the yaml to:

profile::pulp::rpmrepos:
  pulp-2-el7-x86_64:
    display_name: Pulp 2 Production Releases
    feed: https://repos.fedorapeople.org/repos/pulp/pulp/stable/2/7/x86_64/
    relative_url: pulp/stable/2/7/x86_64/
    gpg_key: https://repos.fedorapeople.org/repos/pulp/pulp/GPG-RPM-KEY-pulp-2

Still getting:

[root@mbi-pulp-01 ~]# puppet agent -t
Info: Using configured environment 'production'
Info: Retrieving pluginfacts
Info: Retrieving plugin
Info: Loading facts
Info: Caching catalog for mbi-pulp-01.utmb.lan
Info: Applying configuration version '4de03acbb3fc11e7ecfefea8cf32f1a339105fbd'
Error: Failed to apply catalog: undefined method `each' for nil:NilClass
[root@mbi-pulp-01 ~]#

I also updated my pulp profile recently to include a few more things, one of which trusts the CA cert at the system level, which I read might be an issue in other issues.

class profile::pulp (
  $rpmrepos = {},
  $rpmrepos_defaults = {},
  $ca_cert = 'file:///etc/puppetlabs/puppet/ssl/certs/ca.pem'
) {

  # LVM: DataDisk Mounts - please see hieradata/role/pulp.yaml
  include ::lvm

  #  include ::repos::pulp2
  include ::pulp::repo::upstream

  include ::pulp
  include ::pulp::admin

  include ::ca_cert

  # Populate RPM-GPG-KEY files
  file {'/var/lib/pulp/static/rpm-gpg':
    ensure  => 'directory',
    owner   => 'apache',
    group   => 'apache',
    source  => "puppet:///modules/${module_name}/pulp/rpm-gpg",
    recurse => true,
  }

  # Trust CA Cert
  ca_cert::ca {'PulpCA':
    ensure => 'trusted',
    source => $ca_cert
  }

  # RPM Repos
  create_resources('pulp_rpmrepo', $rpmrepos, $rpmrepos_defaults)

  # Ordering
  Class['lvm'] -> Class['pulp'] -> File['/var/lib/pulp/static/rpm-gpg'] -> Ca_cert::Ca <| |> -> Pulp_rpmrepo <| |> -> Pulp_schedule <| |>

  # Firewall
  firewall { '100 INPUT allow http(s) from all':
    dport  => [ '80', '443' ],
    proto  => 'tcp',
    action => 'accept',
    chain  => 'INPUT',
  }

  firewall { '100 INPUT allow http(s) from all IPv6':
    dport    => [ '80', '443' ],
    proto    => 'tcp',
    action   => 'accept',
    chain    => 'INPUT',
    provider => 'ip6tables',
  }

  firewall { '200 OUTPUT allow http(s) to all':
    dport  => [ '80', '443' ],
    proto  => 'tcp',
    action => 'accept',
    chain  => 'INPUT',
  }

  firewall { '200 OUTPUT allow http(s) to all IPv6':
    dport    => [ '80', '443' ],
    proto    => 'tcp',
    action   => 'accept',
    chain    => 'INPUT',
    provider => 'ip6tables',
  }

  # firewall { '200 OUTPUT allow pulp/qpid ports from all':
  #   dport  => '5672',
  #   proto  => 'tcp',
  #   action => 'accept',
  #   chain  => 'INPUT',
  # }
}

cristifalcas · 2017-05-25T08:58:49Z

Can you remove the gpg_key line and try again? I think the error is from there. The provider tries to read that as a path on the disk.

TJM · 2017-05-25T22:34:50Z

Tried it, same error. Not sure why that would be a nil.each?

For what its worth, my data entries look like:

profile::pulp::rpmrepos:
  centos-7-x86_64-base:
    display_name: CentOS 7 x86_64 Base
    feed: http://mirror.centos.org/centos/7/os/x86_64/
    relative_url: centos/7/os/x86_64/
    gpg_key: /var/lib/pulp/static/rpm-gpg/RPM-GPG-KEY-CentOS-7
    remove_missing: true
    retain_old_count: 2
    serve_http: true
    serve_https: true
    validate: true
    repoview: false

... and I am populating /var/lib/pulp/static/rpm-gpg with puppet. (see manifest above)

TJM · 2017-05-25T22:43:24Z

for what its worth, I ran into a cyclic dependency issue today with this... In order to trust the CA certificate, you install the "ca-certificates" package, which, of course you have to get from pulp, with its untrusted SSL CA ... I had to make base and updates http for now till I figure that one out. :)

~tommy

TJM · 2017-05-25T23:28:17Z

Steps to reproduce

(READY)

start with a base centos 7 machine (DigitalOcean/AWS/etc)
bootstrap.sh (this was provided via user-data originally)

# Variables
hostname='test-pulp-01.test.lan'
role='pulp'
puppet_server='test-puppet-01.test.lan'

# Set hostname (Mostly for AWS)
[ -d /etc/cloud/cloud.cfg.d ] && echo "preserve_hostname: true" > /etc/cloud/cloud.cfg.d/99_hostname.cfg
hostnamectl set-hostname ${hostname}

# Updates and install Puppet Agent
yum -y update
yum -y install https://yum.puppetlabs.com/puppetlabs-release-pc1-el-7.noarch.rpm
yum -y install puppet-agent

## Role
#mkdir -p /etc/puppetlabs/facter/facts.d
#echo "role=$role" > /etc/puppetlabs/facter/facts.d/role.txt
cat > /etc/puppetlabs/puppet/csr_attributes.yaml << YAML
# custom_attributes:
#     1.2.840.113549.1.9.7: mySuperAwesomePassword
extension_requests:
    pp_role: ${role}
YAML

# Configure Puppet Agent
PATH=$PATH:/opt/puppetlabs/puppet/bin
augtool -s "set /files/etc/puppetlabs/puppet/puppet.conf/agent/server ${puppet_server}"
augtool -s "set /files/etc/puppetlabs/puppet/puppet.conf/agent/environment ${puppet_env}"
systemctl enable puppet

echo " *** Don't forget to \"reboot\" to apply updates ***"

reboot # (if updates installed)
In lieu of a proper puppet master, lets generate our own SSL cert, so that everything is in the same place

[root@test-pulp-01 ~]# hostname -f
test-pulp-01.test.lan
[root@test-pulp-01 ~]# puppet cert generate $(hostname -f)
Notice: Signed certificate request for ca
Notice: test-pulp-01.test.lan has a waiting certificate request
Notice: Signed certificate request for test-pulp-01.test.lan
Notice: Removing file Puppet::SSL::CertificateRequest test-pulp-01.test.lan at '/etc/puppetlabs/puppet/ssl/ca/requests/test-pulp-01.test.lan.pem'
Notice: Removing file Puppet::SSL::CertificateRequest test-pulp-01.test.lan at '/etc/puppetlabs/puppet/ssl/certificate_requests/test-pulp-01.test.lan.pem'
[root@test-pulp-01 ~]#

Install pulp module

puppet module install katello-pulp
cd /etc/puppetlabs/code/environments/production/modules
rm -rf pulp
yum -y install git
yum -y install epel-release
- I know I could have done this via puppet, but thats not what we are testing, and might be a bug all by itself.
git clone https://github.com/Katello/puppet-pulp.git pulp

Populate local skeleton of puppet pulp profile and data

mkdir -p /etc/puppetlabs/code/environments/production/modules/profile/manifests
/etc/puppetlabs/code/environments/production/modules/profile/manifests/pulp.pp:

# profile::pulp
class profile::pulp (
  $rpmrepos = {},
  $rpmrepos_defaults = {},
) {
  # LVM: DataDisk Mounts - please see hieradata/role/pulp.yaml
  # include ::lvm
  #  include ::repos::pulp2
  include ::pulp::repo::upstream
  include ::pulp
  include ::pulp::admin

  # RPM Repos
  create_resources('pulp_rpmrepo', $rpmrepos, $rpmrepos_defaults)

  # Ordering
  #Class['lvm'] -> Class['pulp'] -> File['/var/lib/pulp/static/rpm-gpg'] -> Pulp_rpmrepo <| |> -> Pulp_schedule <| |>
  Class['pulp'] -> Pulp_rpmrepo <| |> -> Pulp_schedule <| |>

}

/etc/puppetlabs/code/environments/production/hieradata/common.yaml

---
pulp::debugging_mode: true
pulp::log_level: DEBUG
pulp::ssl_verify_client: optional
pulp::https_cert: /etc/puppetlabs/puppet/ssl/certs/%{::fqdn}.pem
pulp::https_key: /etc/puppetlabs/puppet/ssl/private_keys/%{::fqdn}.pem
pulp::https_chain: /etc/puppetlabs/puppet/ssl/certs/ca.pem

pulp::admin::verify_ssl: false
pulp::admin::logging_filename: /var/log/pulp/admin.log
pulp::admin::logging_call_log_filename: /var/log/pulp/server_calls.log

profile::pulp::rpmrepos_defaults:
  remove_missing: true
  retain_old_count: 2
  serve_http: true
  serve_https: true
  validate: true
  repoview: false

profile::pulp::rpmrepos:
  centos-7-x86_64-base:
    display_name: CentOS 7 x86_64 Base
    feed: http://mirror.centos.org/centos/7/os/x86_64/
    relative_url: centos/7/os/x86_64/
    gpg_key: /etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7
  centos-7-x86_64-updates:
    display_name: CentOS 7 x86_64 Updates
    feed: http://mirror.centos.org/centos/7/updates/x86_64/
    relative_url: centos/7/updates/x86_64/
    gpg_key: /etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7

APPLY!

puppet apply --verbose -e 'include profile::pulp'

RESULTS:

Pulp is installed, but we got this error:

ERROR1:

Error: Failed to apply catalog: https request for repo "# threw exception No such file or directory @ rb_sysopen - /root/.pulp/user-cert.pem. Connection details: url=https://test-pulp-01.test.lan:443/pulp/api/v2/repositories///, cert_path=/root/.pulp/user-cert.pem.

WORKAROUND1:

Get the password from /etc/pulp/server.conf
- grep password /etc/pulp/server.conf
Login to pulp-admin
- pulp-admin login -u admin (paste password when prompted)

[root@test-pulp-01 modules]# grep ^default_password /etc/pulp/server.conf
default_password: (censored)
[root@test-pulp-01 modules]# pulp-admin login -u admin
Enter password:
Successfully logged in. Session certificate will expire at Jun  1 23:44:58 2017
GMT.

[root@test-pulp-01 modules]#
[root@test-pulp-01 modules]# ls -l /root/.pulp/user-cert.pem
-rw------- 1 root root 2120 May 25 23:44 /root/.pulp/user-cert.pem

Run puppet apply again:
- puppet apply --verbose -e 'include profile::pulp'

ERROR2:

Notice: Compiled catalog for test-pulp-01.test.lan in environment production in 2.88 seconds
Info: Applying configuration version '1495756029'
Error: Failed to apply catalog: undefined method `each' for nil:NilClass

Change the rpmrepos to disable creation of repos
- sed -i -e 's/rpmrepos:/DISABLErpmrepos:/' /etc/puppetlabs/code/environments/production/hieradata/common.yaml
Puppet Apply

[root@test-pulp-01 modules]# puppet apply --verbose -e 'include profile::pulp'
Info: Loading facts
Info: Loading facts
Info: Loading facts
Info: Loading facts
Info: Loading facts
Info: Loading facts
##(SNIP DEPRECATION WARNINGS FROM stdlib)###
Notice: Compiled catalog for test-pulp-01.test.lan in environment production in 2.80 seconds
Info: Applying configuration version '1495756252'
Notice: Applied catalog in 1.18 seconds
[root@test-pulp-01 modules]#

I hope this helps!

ekohl · 2017-05-28T20:40:01Z

With the following test.pp:

pulp_rpmrepo { 'centos-7-x86_64-base':
  display_name     => 'CentOS 7 x86_64 Base',
  feed             => 'http://mirror.centos.org/centos/7/os/x86_64/',
  relative_url     => 'centos/7/os/x86_64/',
  gpg_key          => '/etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7',
  remove_missing   => true,
  retain_old_count => 2,
  serve_http       => true,
  serve_https      => true,
  validate         => true,
  repoview         => false,
}

Then it applies succesfully:

[root@centos-7-x64 /]# puppet apply test.pp                  
Notice: Compiled catalog for centos-7-x64.example.com in environment production in 0.08 seconds
Notice: /Stage[main]/Main/Pulp_rpmrepo[centos-7-x86_64-base]/ensure: created
Notice: Applied catalog in 0.36 seconds

I extended the code in #248 with an additional acceptance test locally.

Could you create a repository using pure puppet code? I wonder if there's some odd conversion going on somewhere with hiera.

TJM · 2017-05-30T16:55:48Z

I added "test-" in front, to be sure it wasn't clashing with one of my manually added repos...

  # RPM Repos
  #create_resources('pulp_rpmrepo', $rpmrepos, $rpmrepos_defaults)

  pulp_rpmrepo { 'centos-7-x86_64-base':
    display_name     => 'CentOS 7 x86_64 Base',
    feed             => 'http://mirror.centos.org/centos/7/os/x86_64/',
    relative_url     => 'centos/7/os/x86_64/',
    gpg_key          => '/etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7',
    remove_missing   => true,
    retain_old_count => 2,
    serve_http       => true,
    serve_https      => true,
    validate         => true,
    repoview         => false,
  }

and the results:

[root@mbi-pulp-01 ~]# puppet agent -t
Info: Using configured environment 'production'
Info: Retrieving pluginfacts
Info: Retrieving plugin
Info: Loading facts
Info: Caching catalog for mbi-pulp-01.utmb.lan
Info: Applying configuration version '93cc89a8d5f1b65adcb128f2e485e3e71291f800'
Error: Failed to apply catalog: undefined method `each' for nil:NilClass
[root@mbi-pulp-01 ~]#

still getting the same error :(

I wonder if there is something else (some other packages?) that are interfering somehow?

[root@mbi-pulp-01 ~]# rpm -q puppet-agent
puppet-agent-1.10.1-1.el7.x86_64

Not sure what to check next?

ekohl · 2017-05-30T17:03:54Z

I suspect you have something in your pulp database that the provider doesn't handle properly. I'm currently looking into them and clean them up a bit but for now my knowledge about them is still a bit limited.

TJM · 2017-05-30T17:18:04Z

I see (a/the) problem:

Here is a small portion of the "provider" (that has a .each)

  def self.instances
    all=[]
    @pulp = Puppet::Util::PulpUtil.new
    @pulp.get_repos.each { |repo|
      next if repo['notes']['_repo-type'] != 'rpm-repo'
      hash_properties = get_resource_properties(repo['id'])
      all << new(hash_properties)
    }
    all
  end

irb(main):001:0> require File.expand_path('../../../util/pulp_util', __FILE__)
=> true
irb(main):002:0>     all=[]
=> []
irb(main):003:0>     @pulp = Puppet::Util::PulpUtil.new
=> #<Puppet::Util::PulpUtil:0x00000001448df0 @config={:host=>"mbi-pulp-01.utmb.lan", :port=>"443", :api_prefix=>"/pulp/api", :verify_ssl=>"false", :cert_dir=>"~/.pulp", :cert_file=>"user-cert.pem"}>
irb(main):004:0> @pulp.get_repos
=> nil
irb(main):005

For what its worth, this has plenty of repos, using pulp-admin repo list, but maybe there should be a nil check there? Even if it just emitted a log entry saying it was unable to retrieve repos (which is a different problem)

On that whim, I tried tailing the weblogs and:

==> /var/log/httpd/pulp-https_access_ssl.log <==
172.21.50.219 - admin:admin:5924cd33b96b075501960280 [30/May/2017:17:20:33 +0000] "GET /pulp/api/v2/repositories/ HTTP/1.1" 401 552 "-" "Ruby"
Error: Failed to apply catalog: undefined method `each' for nil:NilClass

ekohl · 2017-05-30T17:22:32Z

That's not the only place where nils can show up and there are no checks for it.

walkamongus · 2017-05-30T17:24:54Z

@TJM the get_repos method doesn't use pulp-admin repo list but rather a direct API call: https://github.com/Katello/puppet-pulp/blob/master/lib/puppet/util/pulp_util.rb#L21

I dunno if this will affect the nil value or not...just noting that there maybe a response difference between the command and the API.

TJM · 2017-05-30T17:45:50Z

I moved my ~/.pulp/admin.conf aside (that had a username and password in it) and tried to use pulp-admin login -u admin ... which said it was successful:

[root@mbi-pulp-01 .pulp]# pulp-admin login -u
==> /var/log/httpd/pulp-https_access_ssl.log <==
172.21.10.49 - mbi-pulp-01 [30/May/2017:17:35:30 +0000] "GET /pulp/api/v2/consumers/mbi-pulp-01/ HTTP/1.1" 401 486 "-" "-"
admin
Enter password:

==> /var/log/httpd/pulp-https_error_ssl.log <==
[Tue May 30 17:35:39.165487 2017] [ssl:warn] [pid 24692] [client 172.21.10.49:33566] AH02227: Failed to set r->user to 'SSL_CLIENT_S_DN_CN'

==> /var/log/httpd/pulp-https_access_ssl.log <==
172.21.10.49 - - [30/May/2017:17:35:39 +0000] "POST /pulp/api/v2/actions/login/ HTTP/1.1" 200 2186 "-" "-"
Successfully logged in. Session certificate will expire at Jun  6 17:35:39 2017
GMT.

[root@mbi-pulp-01 .pulp]#

but...

[1]+  Running                 tail -f /var/log/httpd/*log &
[root@mbi-pulp-01 .pulp]# pulp-admin repo list
+----------------------------------------------------------------------+
                              Repositories
+----------------------------------------------------------------------+

172.21.10.49 - admin:admin:5924cd33b96b075501960280 [30/May/2017:17:39:45 +0000] "GET /pulp/api/v2/repositories/ HTTP/1.1" 401 552 "-" "-"
The specified user does not have permission to execute the given command

So, problems are...

PulpUtil should probably detect a 401 (Unauthorized) and emit some sort of error about it, which would make whatever error handling done in the provider make more sense.
There should be a nil? check on API calls before trying to run .each on them (provider)
My "out of the box" pulp doesn't appear to work with SSL Cert authentication. I did try the API call using curl, and it works with -u "admin:pass", but not --cert ~/.pulp/user-cert.pem (401 unauthorized)

[root@mbi-pulp-01 ~]# curl  -cert ~/.pulp/user-cert.pem -s -D - https://mbi-pulp-01.utmb.lan/pulp/api/v2/repositories/
HTTP/1.1 401 UNAUTHORIZED
Date: Tue, 30 May 2017 18:21:55 GMT
Server: Apache/2.4.6 (CentOS)
Content-Length: 417
Content-Type: application/json; charset=utf-8

{"http_request_method": "GET", "exception": null, "error_message": "Authentication with username None failed: invalid SSL certificate.", "_href": "/pulp/api/v2/repositories/", "http_status": 401, "error": {"code": "PLP0027", "data": {"user": null}, "description": "Authentication with username None failed: invalid SSL certificate.", "sub_errors": []}, "traceback": null, "auth_error_code": "invalid_ssl_certificate"

The one thing that might be different then... according to my puppet manifest, I am using "repos::upstream" ... maybe I should not be?

TJM · 2017-05-30T18:39:22Z

Updating the curl line from above, --cert tommy :)

[root@mbi-pulp-01 .pulp]# curl  --cert ~/.pulp/user-cert.pem -s -D - https://mbi-pulp-01.utmb.lan/pulp/api/v2/repositories/
HTTP/1.1 401 UNAUTHORIZED
Date: Tue, 30 May 2017 18:38:20 GMT
Server: Apache/2.4.6 (CentOS)
Content-Length: 552
Content-Type: application/json; charset=utf-8

{"http_request_method": "GET", "exception": null, "error_message": "Authentication with username admin:admin:5924cd33b96b075501960280 failed: preauthenticated remote user is missing.", "_href": "/pulp/api/v2/repositories/", "http_status": 401, "error": {"code": "PLP0029", "data": {"user": "admin:admin:5924cd33b96b075501960280"}, "description": "Authentication with username admin:admin:5924cd33b96b075501960280 failed: preauthenticated remote user is missing.", "sub_errors": []}, "traceback": null, "auth_error_code": "pre_auth_remote_user_missing"}

TJM · 2017-05-30T18:50:59Z

https://www.redhat.com/archives/pulp-list/2016-March/msg00050.html

I took the SSLUsername SSL_CLIENT_S_DN_CN out of the config, and its happier now....

https://github.com/Katello/puppet-pulp/blob/master/manifests/apache.pp#L51-L62

https://github.com/Katello/puppet-pulp/blob/master/manifests/init.pp#L105-L106

I think I can workaround this by setting pulp::ssl_username: false in hieradata. I wonder if that "default" value needs adjusted?

TJM · 2017-05-30T19:28:43Z

OK, so I have solved Problem 3 (my SSL Cert auth/e was broken), but these are still up for grabs:

PulpUtil should probably detect a 401 (Unauthorized) and emit some sort of error about it, which would make whatever error handling done in the provider make more sense.
There should be a nil? check on API calls before trying to run .each on them (provider)

ekohl · 2017-05-31T07:41:40Z

I think I can workaround this by setting pulp::ssl_username: false in hieradata. I wonder if that "default" value needs adjusted?

I think it does. See #138 as well for why it's there but it's not a good default.

I wasn't aware (yet) that it does a direct API call. That makes me wonder if we should drop pulp-admin altogether and fully move to the API.

walkamongus · 2017-05-31T12:13:25Z

I would say either use the API or use pulp-admin for everything. I have a pulp rpm repo provider written that uses only API calls if anyone is interested. I am requiring the Puppet user to set login credentials in a file and these are used to login and retrieve certificates via the API whenever the user cert is expired.

TJM · 2017-05-31T15:41:31Z

@ekohl - Yes, #138 is the same issue I saw. The problem is the way that pulp server generates its auth certs is not compatible with whatever else is generating auth certs (guessing thats katello related).... and I do realize that I am using Katello/puppet-pulp, but I would say the default needs to change.

I agree with @walkamongus. I suggest going all one direction or the other. The pulp-admin command needs the ability to output "structured" data (json/xml/yaml), which would eliminate the need to write an API wrapper, but for now, fixing the API wrapper (pulp::util) to detect error conditions (maybe using yours) would be a good interim fix.

Tommy

ccnifo · 2017-06-02T09:31:53Z

I would say either use the API or use pulp-admin for everything.

I quite agree with that but it looks like each one is used when it's more relevant: state checking is done through the API (whose syntax fits better for that) and modifications are made much easier by using pulp-admin, which makes adding optional params easy.

TJM changed the title ~~unable to add pulp_rpmrepo~~ unable to add pulp_rpmrepo (undefined method `each' for nil:NilClass) May 25, 2017

TJM mentioned this issue May 27, 2017

How to use Pulp - need docs #241

Closed

ekohl added the Waiting on contributor label May 29, 2017

TJM mentioned this issue May 31, 2017

Authentification issue #206

Closed

ekohl mentioned this issue Jun 1, 2017

Add basic provider for Pulp roles #234

Merged

ekohl mentioned this issue Jul 12, 2017

Clean up types and providers #252

Merged

ekohl closed this as completed in #252 Jul 12, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

unable to add pulp_rpmrepo (undefined method `each' for nil:NilClass) #245

unable to add pulp_rpmrepo (undefined method `each' for nil:NilClass) #245

TJM commented May 24, 2017

ekohl commented May 24, 2017

TJM commented May 25, 2017

cristifalcas commented May 25, 2017

TJM commented May 25, 2017 •

edited

Loading

TJM commented May 25, 2017

TJM commented May 25, 2017 •

edited

Loading

ekohl commented May 28, 2017

TJM commented May 30, 2017

ekohl commented May 30, 2017

TJM commented May 30, 2017 •

edited

Loading

ekohl commented May 30, 2017

walkamongus commented May 30, 2017

TJM commented May 30, 2017 •

edited

Loading

TJM commented May 30, 2017 •

edited

Loading

TJM commented May 30, 2017

TJM commented May 30, 2017

ekohl commented May 31, 2017

walkamongus commented May 31, 2017

TJM commented May 31, 2017

ccnifo commented Jun 2, 2017

unable to add pulp_rpmrepo (undefined method `each' for nil:NilClass) #245

unable to add pulp_rpmrepo (undefined method `each' for nil:NilClass) #245

Comments

TJM commented May 24, 2017

ekohl commented May 24, 2017

TJM commented May 25, 2017

cristifalcas commented May 25, 2017

TJM commented May 25, 2017 • edited Loading

TJM commented May 25, 2017

TJM commented May 25, 2017 • edited Loading

Steps to reproduce

RESULTS:

ERROR1:

WORKAROUND1:

ERROR2:

ekohl commented May 28, 2017

TJM commented May 30, 2017

ekohl commented May 30, 2017

TJM commented May 30, 2017 • edited Loading

ekohl commented May 30, 2017

walkamongus commented May 30, 2017

TJM commented May 30, 2017 • edited Loading

TJM commented May 30, 2017 • edited Loading

TJM commented May 30, 2017

TJM commented May 30, 2017

ekohl commented May 31, 2017

walkamongus commented May 31, 2017

TJM commented May 31, 2017

ccnifo commented Jun 2, 2017

TJM commented May 25, 2017 •

edited

Loading

TJM commented May 25, 2017 •

edited

Loading

TJM commented May 30, 2017 •

edited

Loading

TJM commented May 30, 2017 •

edited

Loading

TJM commented May 30, 2017 •

edited

Loading