neutron

I keep hitting the same brick wall with Quantum

Asked by Joshua Dotson

Hi,

I've pursued a GRE-enabled 3-node Folsom setup with a provider router and per-tenant networks for a long time. However, every time I get to the end of the guides, all of them, regardless of 12.04 LTS or 12.10, regardless of scouring every log on all three nodes for answer, it simply doesn't work as described.

I feel there must a major bug in the code, if all of the guides which worked, when the authors penned them, no longer work. If you look at all of the guides for this, namespaces and overlapping IP's are disabled. Yet, in a recent ticket, both myself and a completely separate person experienced the same need for namespaces to be enabled. (https://answers.launchpad.net/quantum/+question/216939)

Can anyone confirm that this setup REQUIRES namespaces? I'd very much like to retain metadata and security groups. The OVS interfaces on the network node never come up, when they are disabled. Surely, all the guides on the topic aren't wrong. The only guide not recommending disabled namespaces is Cisco's: http://docwiki.cisco.com/wiki/Cisco_OpenStack_Edition:_Folsom_Manual_Install

If namespaces are required for this setup, why? Is it the conflict of having the DHCP and L3 agents on the same node, as noted in the docs? If so, can I put on or the other on the Control node to clean this up? Could I introduce a fourth node to do one of the two agents?

I would very much like to troubleshoot this issue and see the official guides actually work. At the moment though, I'm more worried about my OS environment working, than seeing the documentation working.

Thanks, and please pardon the frustration. :-) I mean well and fully grant this could be habitual user error.

Joshua

Question information

Language:
English Edit question
Status:
Open
For:
neutron Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Joshua Dotson (tns9) said : 		#1

I moved the DHCP agent to the compute node to separate it from the L3 agent and hopefully stem any conflict. I uninstalled it on the network node. The issue above persists, even after this change. The tap interfaces on both network and compute are down, per ifconfig. It's almost like namespaces are half enabled, but they are not.

root@khyp-c49x:/var/lib/quantum/dhcp/1c42e215-30fb-4413-a596-ce6e3d99addf# cat *
fa:16:3e:58:02:99,10-5-5-1.openstacklocal,10.5.5.1
fa:16:3e:6b:e1:dc,10-5-5-3.openstacklocal,10.5.5.3
fa:16:3e:70:7d:57,10-5-5-2.openstacklocal,10.5.5.2
fa:16:3e:18:4a:be,10-5-5-4.openstacklocal,10.5.5.4
tapf16436e3-f6tag:tag0,option:dns-server,192.168.10.88
tag:tag0,option:router,10.5.5.12772
root@khyp-c49x:/var/lib/quantum/dhcp/1c42e215-30fb-4413-a596-ce6e3d99addf# ifconfig tapf16436e3-f6
tapf16436e3-f6 Link encap:Ethernet HWaddr 56:88:7a:8a:a8:e3
          inet addr:10.5.5.4 Bcast:10.5.5.255 Mask:255.255.255.0
          BROADCAST MULTICAST MTU:1500 Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)

root@khyp-c49x:/var/lib/quantum/dhcp/1c42e215-30fb-4413-a596-ce6e3d99addf# tcpdump -i tapf16436e3-f6
tcpdump: tapf16436e3-f6: That device is not up
root@khyp-c49x:/var/lib/quantum/dhcp/1c42e215-30fb-4413-a596-ce6e3d99addf#
root@khyp-c49x:/var/lib/quantum/dhcp/1c42e215-30fb-4413-a596-ce6e3d99addf#

Revision history for this message
Joshua Dotson (tns9) said : 		#2

Is there anything I can provide to help? I'm willing to do any number of tests. As my deadlines loom, I'm planning a retreat to nova-network, if I cannot make sense of this topic.

Thank you,
Joshua

Revision history for this message
Thiago Martins (martinx) said : 		#3

BTW, for God's sake!!

Why the Ubuntu Openstack does not create its SQL databases automatically?

Why it does not use Debconf and db-config???

Why there is so many manual steps?!

Why we can't have: "apt-get install openstack-frontent", "apt-get install openstack-network" and "apt-get install openstack-compute" fully automated?

The Openstack Ubuntu package maintainers needs to understand that this current implementation isn't designed for humans. This thing is light years away from what Ubuntu proposes (motto).

The Ubuntu oficial documentation located here: https://help.ubuntu.com/12.04/serverguide/ubuntucloud.html is creepy! This isn't professional. You guys must learn a lot before creating a distro for humans.

My condolences... For a dead procedure...

Sorry Joshua, I'm trying to deploy Openstack about a month now, without any kind of success. I would love to help you but... I can't even help myself now... lol

Anyway, I can't give it up, since this is my job and I need to deploy Openstack without ANY error. But, there is no documentation that we can follow.

Best,
Thiago

Revision history for this message
Thiago Martins (martinx) said : 		#4

Also, the documentation that does NOT work is this: http://docs.openstack.org/folsom/basic-install/content/basic-install_intro.html

Revision history for this message
Thiago Martins (martinx) said : 		#5

Maybe this is related:

https://bugs.launchpad.net/ubuntu/+source/nova/+bug/1089791

https://answers.launchpad.net/quantum/+question/211021
https://answers.launchpad.net/quantum/+question/212874

No IPs for my Instance, my Instance Logs shows:

----------
cloud-init start-local running: Thu, 03 Jan 2013 04:20:52 +0000. up 8.53 seconds

no instance data found in start-local

cloud-init-nonet waiting 120 seconds for a network device.

cloud-init-nonet gave up waiting for a network device.

ci-info: lo : 1 127.0.0.1 255.0.0.0 .

ci-info: eth0 : 1 . . fa:16:3e:83:06:4f

route_info failed

Waiting for network configuration...

Waiting up to 60 more seconds for network configuration...

Booting system without full network configuration...
----------

I really appreciate ANY help!

Thanks!
Thiago

Revision history for this message
Stephen (stephen-kahmann) said : 		#6

I am also having many issues with quantum outlined here: <https://answers.launchpad.net/quantum/+question/218960>. I really want to get this environment up and running, but for some reason the services aren't able to bring up the bridge interfaces.

Revision history for this message
Spatialist (fsluiter) said : 		#7

Hi,
I ran into a major issue when just installing openstack on multiple nodes:
we have an NFS share, and it of course needs users to have the same uid on the whole cluster.
By default apt-get picks the first free uid. And therefore the uids were different between the nodes!
As error messages are cryptic, I only found this out after days!

You can check this by comparing /etc/passwd on all the nodes in your cluster.
If the uids and groups are different, changes are you have a problem.

Revision history for this message
Joshua Dotson (tns9) said : 		#8

Spatialist: I'm sorry, but that is not the case with my installs. I am not using NFS.

Thanks,
Joshua

Revision history for this message
Spatialist (fsluiter) said : 		#9

Joshua: it might also cause problems without the NFS.
Another silly thing was that RabbitMQ has a default user guest with password guest. Some applications have this hardcoded as default (nova), but for some crazy reason, quantum has another default in the documentation.
Now when you follow the documentation, quantum cannot acces RabbitMQ.

Good luck

Revision history for this message
Spatialist (fsluiter) said : 		#10

oops, wrong bug commented. Sorry

Revision history for this message
Joshua Dotson (tns9) said : 		#11

Thanks, but that is not the issue here.

-Joshua

Can you help with this problem?

Provide an answer of your own, or ask Joshua Dotson for more information if necessary.

To post a message you must log in.