Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 890841

Summary: Openstack Installer: packstack misconfigure nova.conf and api-paste.ini values with 127.0.0.1
Product: Red Hat OpenStack Reporter: Nir Magnezi <nmagnezi>
Component: openstack-packstackAssignee: Derek Higgins <derekh>
Status: CLOSED DUPLICATE QA Contact: Nir Magnezi <nmagnezi>
Severity: high Docs Contact:
Priority: unspecified    
Version: 2.0 (Folsom)CC: aortega, derekh, ykaul
Target Milestone: ---Keywords: Triaged
Target Release: ---   
Hardware: Unspecified   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2013-01-19 09:28:23 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
packstack log (DEBUG mode)
none
answers file none

Description Nir Magnezi 2012-12-30 15:46:42 UTC 
Description of problem:
=======================
Used packstack to install openstack as follows:

Cloud Controller installed with: dashboard, cinder, nova-scheduler, database, queue, nova-cert, nova-consoleauth, glance-api, glance-registry, keystone
Nova compute node installed with: nova-compute, nova-network, nova-api

packstack fails with an error.


Version-Release number of selected component (if applicable):
============================================================
Folsom.
Packstack Ver: openstack-packstack-2012.2.2-0.1.dev262.el6.noarch


How reproducible:
=================
100%


Steps to Reproduce:
===================
1. Use packstack with the answer file attached.
   Make sure you run packstack in the Cloud Controller CLI.


 
Actual results:
===============
1. packstack exit with the following error:

Testing if puppet apply is finished : 10.35.110.15_api_nova.pp.log                          [ ERROR ]
Error during puppet run : err: /Stage[main]/Nova::Api/Exec[nova-db-sync]: Failed to call refresh: /usr/bin/nova-manage db sync returned 1 instead of one of [0] at /var/tmp/a1817a9e-cf20-465a-b241-9f1009f7c05b/modules/nova/manifests/api.pp:98
Please check log file /var/tmp/a1817a9e-cf20-465a-b241-9f1009f7c05b/openstack-setup_2012_12_30_17_12_41.log for more information

I've checked the configuration files on nova compute node.
packstack misconfigured nova.conf with 127.0.0.1, While it should have set the routable Cloud Controller IP Address:

glance_api_servers=127.0.0.1:9292
sql_connection=mysql://nova:nova_default_password.0.1/nova
qpid_hostname=127.0.0.1

Same goes for api-paste.ini:
auth_host=127.0.0.1

* Please note that 127.0.0.1 is the default value set by packstack in the answers file.
  I ran packstack from the Cloud Controller machine, Therefore this value should be valid.
  In order to support this topology, packstack should "translated" the loopback address to a routable IP Address.


Expected results:
=================
packstack should run with no errors.
Comment 1 Nir Magnezi 2012-12-30 15:49:52 UTC 
Created attachment 670434 [details]
packstack log (DEBUG mode)
Comment 2 Nir Magnezi 2012-12-30 15:50:37 UTC 
Created attachment 670435 [details]
answers file
Comment 4 Yaniv Kaul 2012-12-31 08:57:56 UTC 
I think the logic should be (warning, pseudo-code)
 if any of the IPs != 127.0.0.1, then none of the IPs should be 127.0.0.1
Comment 5 Perry Myers 2012-12-31 12:10:12 UTC 
(In reply to comment #4)
> I think the logic should be (warning, pseudo-code)
>  if any of the IPs != 127.0.0.1, then none of the IPs should be 127.0.0.1

Agree and marking this bug as Triaged as I just ran into this same issue.  Basically, 127.0.0.1 should only be valid if you are doing an 'all in one' install.  If you have any multi-host at all, setting anything to 127.0.0.1 will either cause packstack to fail (as above) or just result in a non-working configuration.

Perhaps we should provide a cmdline option/config file param for all-in-one that sets the IP addr in that case for all services to 127.0.0.1, but unless you set this specific option/flag then it never defaults to that, to prevent you from accidentally selecting it in a multi-host install scenario?
Comment 6 Nir Magnezi 2012-12-31 12:40:53 UTC 
(In reply to comment #5)
> (In reply to comment #4)
> > I think the logic should be (warning, pseudo-code)
> >  if any of the IPs != 127.0.0.1, then none of the IPs should be 127.0.0.1
> 
> Agree and marking this bug as Triaged as I just ran into this same issue. 
> Basically, 127.0.0.1 should only be valid if you are doing an 'all in one'
> install.  If you have any multi-host at all, setting anything to 127.0.0.1
> will either cause packstack to fail (as above) or just result in a
> non-working configuration.
> 
> Perhaps we should provide a cmdline option/config file param for all-in-one
> that sets the IP addr in that case for all services to 127.0.0.1, but unless
> you set this specific option/flag then it never defaults to that, to prevent
> you from accidentally selecting it in a multi-host install scenario?

After further consultation with Yaniv Kaul, I think that we should avoid from 127.0.0.1 even in all-in-one installations.
If we do use it, The user will have to modify almost all .conf files in order to expand his system.
Comment 7 Derek Higgins 2013-01-17 08:07:35 UTC 
patch to change packstack to no longer default to 127.0.0.1
https://review.openstack.org/#/c/19867/

this will prevent this problem occurring
Comment 8 Derek Higgins 2013-01-19 09:28:23 UTC 
This is a dup of
https://bugzilla.redhat.com/show_bug.cgi?id=886541

which is now ON_QA

*** This bug has been marked as a duplicate of bug 886541 ***