3.3. Capsule Issues

This section describes known issues with the Capsule component of Satellite 6.1.
BZ 1199992: Error message for failed content view publishing is displayed in other language
Issue

When you publish a Content View, you might see an error message displayed in a language different from your preferred setting. For example: 

"ERF12-4115 [ProxyAPI::ProxyException]: Klassen für KT_The_Demo_Org_Auto_Library_small_cv_22  konnten nicht von Puppet geladen werden ([RestClient::Forbidden]: 403 Forbidden) für Proxy https://satellite.example.com:9090/puppet"
This can occur if the services with the Satellite server cannot verify the host name being used for the API calls being used.
Workaround

Check /etc/foreman-proxy/settings.yml file and the :trusted_hosts: field. Ensure that the host name in that field matches the host name from the error message above. If not, add an additional entry to that setting (can be multiple lines) and restart the foreman-proxy service. Navigate to Monitor → Tasks and resume the Publish task.

BZ 1205840: Capsule and Pulp might return many "Resetting dropped connection" messages during synchronization
Issue

After you configure and register a capsule to a Satellite server and start the synchronization process, you might see a large number of "Resetting dropped connection" messages in the /var/log/messages file. For example: 

Mar 25 19:09:32 ibm-x3550m3 pulp: requests.packages.urllib3.connectionpool:INFO: Resetting dropped connection: rhsm.example.com
The synchronization process does eventually complete, although the overall process might take longer as a result of this issue. Refer to the bugzilla report for more information.
Workaround

To work around this issue, set the value of MaxRequestsPerChild to 0 in the /etc/httpd/conf.d/prefork.conf file on the Satellite server. This effectively allows an unlimited number of connections to the Satellite server during a Capsule synchronization.

BZ 1117984: Installation program fails if --certs-tar value is incorrect
Issue

If the file name entered for --certs-tar is not correct, the installation program will commence and eventually fail for confusing reasons. The error message may appear as: 

Something went wrong! Check the log for ERROR-level output
  The full log is at /var/log/katello-installer/capsule-installer.log

----

[ INFO 2014-10-23 18:10:41 verbose] All hooks in group post finished
[ERROR 2014-10-23 18:10:41 verbose] Repeating errors encountered during run:
[ERROR 2014-10-23 18:10:41 verbose]  tar -xzf /capsule1.example.com.tar returned 2 instead of one of [0]
[ERROR 2014-10-23 18:10:41 verbose]  /Stage[main]/Capsule/Certs::Tar_extract[/capsule1.example.com.tar]/Exec[extract /capsule1.example.com.tar]/returns: change from notrun to 0 failed: tar -xzf /capsule1.example.com.tar returned 2 instead of one of [0]
Workaround

You need to ensure that the --certs-tar value in the capsule-installer command refers to the correct tar archive. If not, the error condition may be difficult to diagnose.

BZ 1214240: The subscription-manager always shows "Your identity certificate has expired" when system registered through Capsule
Issue

If the time difference between the Satellite and the Capsule is too great, you might see the following error when you try to attach subscriptions. This can occur if the time difference is more than one or two minutes. 

# subscription-manager attach --auto
Your identity certificate has expired
Workaround

The time settings for both the Satellite and Capsule need to be the same to ensure proper registration of clients through the Capsule. Red Hat recommends that you install ntpd or a similar service on the Satellite and Capsule servers, and ensure the times are synchronized. Alternatively, you can manually ensure the time settings on each server are set correctly.

BZ 1207972: Deleting a Capsule can fail with foreign key constraint violation
Issue

If you attempt to delete a Capsule that still has Content Hosts associated with it, the following error appears: 

Operation FAILED: ERROR:  update or delete on table "smart_proxies" violates foreign key constraint "hosts_puppet_ca_proxy_id_fk" on table "hosts"
DETAIL:  Key (id)=(2) is still referenced from table "hosts".
Workaround

The workaround for this issue involves several steps, and also requires that the user deleting the capsule have access to the Organization associated with the Capsule.

Procedure 3.1. To Remove a Capsule:

  1. Remove all hosts provisioned by the Capsule that you want to remove.
  2. Remove the association between the Capsule and subnet.
  3. Remove the association between the Capsule and host group.
  4. Unregister the Capsule.
  5. Delete the Capsule.
BZ 1228165: Possible to delete Content Host for a Capsule, which breaks synchronization
Issue

Deleting or unregistering a Content Host that is associated with a Capsule, and then reregistering it, can cause synchronization to fail with a "Content Host not found" error. For example: 

# hammer -u admin -p changeme capsule content synchronize --id=2
Could not synchronize capsule content:
  Could not find Content Host with exact name 'sat6-capsule-rhel66.example.com', verify the Capsule is registered with that name.
Workaround

You can fix this issue in the foreman-rake console as follows: 

User.current = User.anonymous_admin

SmartProxy.all.each do |proxy|
  content_host = ::Katello::System.where(:name => proxy.name).order("created_at DESC").first

  if content_host
    proxy.content_host_id = content_host.id
    proxy.save!
    puts "Capsule #{proxy} has Content Host #{content_host}"
  else
    puts "No Content Host Found for #{proxy}"
  end
end
This issue has also been observed during an apparently successful upgrade from 6.0.8 to 6.1, but where the migration failed silently. The migration is performed by the db/migrate/20150423134004_add_content_host_id_to_smart_proxy.rb task, which is designed to add the Content Host to the Capsule. If that task fails, the error described here can occur.