Project

General

Profile

Actions

Bug #6484

closed

qpidd produces errors in /var/log/messages when previously subscribed clients are left pointing at a rebuilt host

Added by Roy Williams about 10 years ago. Updated about 6 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
-
Category:
Client/Agent
Target version:
Difficulty:
hard
Triaged:
Fixed in Releases:
Found in Releases:

Description

Problem:
A barrage of errors occur when client machines that were registered to a hostname, for example engr-lbs.uncc.edu, then that host is rebuilt causing it to have new certificates that differ from what the clients were originally registered with to engr-lbs.uncc.edu. This will cause "pulp: kombu.transport.qpid:ERROR: connection aborted" errors to appear in rapid succession approximately every minute, and causes other issues to occur like dynflow becoming stuck at planned state.

Reproduction:
1. Install RHEL 6.5 onto a host.
2. Use subscription-manager to register and subscribe the host.
3. Update the host via "yum update -y"
4. Disable selinux via /etc/sysconfig/selinux by changing SELINUX line to SELINUX=disabled
5. Reboot
6. Install ruby and git, "yum install ruby git -y"
7. Run: git clone https://github.com/Katello/katello-deploy.git
8. Run: cd katello-deploy/
9. Run: ./setup.rb rhel6
10. Run: service iptables disable
11. Register clients to the machine, I have had 15 registered to mine from previous occasions.
12. Do Steps 1-10 again and you will see "pulp: kombu.transport.qpid:ERROR: connection aborted" occurring related to the total number of host pointed at the machine

Actions #1

Updated by Roy Williams about 10 years ago

To add the clients I registered were RHEL 7.0 GA

Actions #3

Updated by Roy Williams about 10 years ago

Another error that occurs post installation before things start getting crazy with the kombu.tarnsport.qpid issue.

"qpidd46470: 2014-07-03 09:24:55 [System] error Error reading socket: Success(0)"

Actions #4

Updated by Eric Helms about 10 years ago

  • Category set to Client/Agent
  • Target version set to 49
  • Difficulty set to hard
  • Triaged changed from No to Yes
Actions #5

Updated by Roy Williams about 10 years ago

I can temporarily resolve the issue via:
Changing the firewall to block port 5671/tcp then running:
foreman-rake jobs:clear

This seems to let Satellite recover from the issue, as long as those host remain blocked.

Actions #6

Updated by Roy Williams about 10 years ago

To add the clients are registered via rhsm and goferd/katello-agent.

Actions #7

Updated by Eric Helms about 10 years ago

  • Target version changed from 49 to 54
Actions #8

Updated by Eric Helms about 10 years ago

  • Target version deleted (54)
Actions #9

Updated by Eric Helms over 8 years ago

  • Translation missing: en.field_release set to 86
Actions #10

Updated by Eric Helms over 8 years ago

  • Status changed from New to Resolved

A lot of change has occurred, so I am moving this to resolved. Please re-open if you continue hitting this issue.

Actions

Also available in: Atom PDF