During a Satellite 6.2 upgrade, as part of the host unification feature, if a host and content host are found with same name but within different organizations, the content host will be unregistered. This will remove the content host from the Satellite. Consequently, once the content host has been unregistered by the upgrade process, it is possible that errors will be observed in /var/log/messages as the content host may still attempt to interact with the Satellite via the agent.
The errors can include ones similar to the following:
[Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Workaround
Prior to performing the Satellite 6.2 upgrade, the user should execute the pre-upgrade script on the Satellite 6.1 Server to obtain a summary of the content hosts that will be unregistered. The script may be executed as:
foreman-rake katello:upgrade_check
Prior to the upgrade, the user should then update the hosts and content hosts to ensure that they are in the correct organizations. If this is not done prior to the upgrade, the user will need to re-register each of those content hosts using the Red Hat Subscription Manager command-line tool after the upgrade has completed.
Description of problem:
I was trying upgrade from sat6.1.9 -> sat6.2 snap15 on rhel7, and installer was throwing following error message in /var/log/messages:
==> /var/log/messages <==
Jun 9 03:51:04 qe-sat6-upgrade-rhel7 qpidd: 2016-06-09 03:51:04 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:51:04 qe-sat6-upgrade-rhel7 qpidd[9999]: 2016-06-09 03:51:04 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:51:46 qe-sat6-upgrade-rhel7 qpidd: 2016-06-09 03:51:46 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:51:46 qe-sat6-upgrade-rhel7 qpidd[9999]: 2016-06-09 03:51:46 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:52:27 qe-sat6-upgrade-rhel7 qpidd: 2016-06-09 03:52:27 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:52:27 qe-sat6-upgrade-rhel7 qpidd[9999]: 2016-06-09 03:52:27 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:53:09 qe-sat6-upgrade-rhel7 qpidd[9999]: 2016-06-09 03:53:09 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:53:09 qe-sat6-upgrade-rhel7 qpidd: 2016-06-09 03:53:09 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Jun 9 03:53:50 qe-sat6-upgrade-rhel7 qpidd[9999]: 2016-06-09 03:53:50 [Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Version-Release number of selected component (if applicable):
upgrade from sat6.1.9 -> sat6.2 snap15 on rhel7,
How reproducible:
Steps to Reproduce:
0 set 6.2 repo
1. yum update -y
2. satellite-installer --scenario satellite --upgrade
Actual results:
[Protocol] error Error on attach: Node not found: pulp.agent.737d5dde-2ec8-4b21-9933-3e7fa45cbf77
Expected results:
no error message should be thrown in logs while installer is running
Additional info:
Sachin, is this error from SNAP 15 or 15.1?
Neither myself or Chris are seeing the error; however, we do have different data and we are running on 15.1.
In discussing this issue with Justin and Jeff, it appears that the issue is due to the fact that the during the upgrade some of the content hosts (pulp consumers) can be deleted from the server. Unfortunately, the agent running on the client (content host) does not know that it has been deleted and it may still reach out to Satellite triggering the error to be generated in the Satellite logs. Once rhsmcertd runs on the client, it will detect that it is no longer registered and stop sending the requests to the server.
This behavior exists today in Satellite 6.1 as well. Example scenario would be, client registers, goferd is running on the client and the content host is deleted from the UI.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
https://access.redhat.com/errata/RHBA-2016:1501