Ask Your Question
2

Devstack install - Gateway is not valid on subnet

asked 2014-08-08 02:51:48 -0500

gepetto gravatar image

updated 2014-08-08 15:27:26 -0500

smaffulli gravatar image

I am working my way through the installation of Devstack on Centos 6.5 and have a network configuration issue. This is a test system and I can have my way with it. My physical network is as follows:

10.0.0.1 - router / gateway to outside world, <br>
10.0.0.2 - 99 - not used, <br>
10.0.0.100 - 200 - dhcp address pool provided by router / gateway, <br>
10.0.0.117 - dhcp address provided to Centos 6.5 system, <br>
10.0.0.201 - 223 - not used, <br>
10.0.0.224 - 255 - Selected for FLOATING_RANGE

I have made the following changes to local.conf:

HOST_IP=10.0.0.117       <--  this is the Centos 6.5 machine eth0<br>
FLOATING_RANGE=10.0.0.224/27  <-- unused ip addresses on the existing network<br>
FIXED_RANGE=172.16.0.0/24  <-- this is what I am hoping will be my virtual network<br>
FIXED_NETWORK_SIZE=256<br>
PUBLIC_INTERFACE=eth0<br>
FLAT_INTERFACE=eth0<br>

I ssh into the machine from a tablet execute ./stack.sh & as the user where I performed the git, etc. Amazing things happen for a while... and then the process terminates with the following error:

[ERROR] /home/gepetto/devstack/functions-common:433 Failure creating SUBNET_ID for 265da70366a142da9357d97af5ab254c<br>

Above the error in the log I see the following statement:

neutron subnet-create --tenant-id 265da70366a142da9357d97af5ab254c --ip_version 4 --gateway 10.0.0.1 --name private-subnet 2cd403ce-70b5-406d-8bf8-32b0f44a1dd0 172.16.0.0/24<br>

I'm not sure how to resolve my configuration issues at this point. Is it possible that I have a conflict between the existing physical router / gateway (also 10.0.0.1) and the virtual one being created? A possible dhcp conflict between the hardware and q-dhcp? I'm not sure what to try next. Do I understand the network parameters? Too much info, not enough info, what did I forget? Below are the last 13 lines of stack.sh.log.

This was a fresh install of Centos 6.5, updates applied, and Devstack installed. As I make my way through my errors, I execute clean.sh, reboot, make changes to local.conf, and re-execute stack.sh.

2014-08-08 06:50:37.120 | ++ neutron subnet-create --tenant-id 265da70366a142da9357d97af5ab254c --ip_version 4 --gateway 10.0.0.1 --name private-subnet 2cd403ce-70b5-406d-8bf8-32b0f44a1dd0 172.16.0.0/24<br>
2014-08-08 06:50:37.120 | ++ read data<br>
2014-08-08 06:50:37.123 | ++ grep ' id '<br>
2014-08-08 06:50:38.956 | Invalid input for operation: Gateway is not valid on subnet. (HTTP 400) (Request-ID: req-d00c4f0f-2518-47fe-8d96-6ec2f45f4370)<br>
2014-08-08 06:50:39.030 | + SUBNET_ID=<br>
2014-08-08 06:50:39.030 | + die_if_not_set 433 SUBNET_ID 'Failure creating SUBNET_ID for 265da70366a142da9357d97af5ab254c'<br>
2014-08-08 06:50:39.030 | + local exitcode=0<br>
2014-08-08 06:50:39.037 | [Call Trace]<br>
2014-08-08 06:50:39.037 | ./stack.sh:1231:create_neutron_initial_network<br>
2014-08-08 06:50:39.037 | /home/gepetto/devstack/lib/neutron:433:die_if_not_set<br>
2014-08-08 06:50:39.037 | /home/gepetto/devstack/functions-common:251:die<br>
2014-08-08 06:50:39.044 | [ERROR] /home/gepetto ...
(more)
edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2014-08-15 01:05:31 -0500

gepetto gravatar image

After much frustration I was able to get past the error. Attempts were made with multiple VBox guests, both Centos and Ubuntu - Actually thought I had it on the Ubuntu 12.04 VM since it appeared to fail on a timeout waiting for c-api to start. ps -ef revealed that cinder had indeed started. Thought that failure was due to the system being a VM and me probably doing something in the host system that caused a delay in the VM. I swapped out the hard drive and did a fresh install of UB 12.04 to give it both processors and all of the memory. I kept thinking about one page I saw that said 'just follow the three steps...' Anyway, I copied in the local.conf used in the UB12.04 devstack install, modified the local.conf to reflect the current IP address, ssh'd in from a tablet and gave ./stack.sh a go. No joy again, but it failed with the error above and not the timeout. Then I started commenting out lines in the local.conf. First to go were PUBLIC_INTERFACE and FLAT_INTERFACE (let the script find them...). Still no joy.

Next to go was a section I found that said add this to enable Neutron on a single node:

# enable - In order to enable Neutron a single node setup, you'll need the
#          following settings in your `local.conf`:

    #disable_service n-net
    #enable_service q-svc
    #enable_service q-agt
    #enable_service q-dhcp
    #enable_service q-l3
    #enable_service q-meta
    #enable_service q-metering
    # Optional, to enable tempest configuration as part of DevStack
    #enable_service tempest

FINALLY!! The build / install completed! I can log in and access the dashboard via the browser on the local machine. Looking around everything looks reasonable, but then again I really don't know what I am doing.

Here are the changes I made / kept in local.conf:

SERVICE_TOKEN=password
ADMIN_PASSWORD=password
MYSQL_PASSWORD=password
RABBIT_PASSWORD=password
SERVICE_PASSWORD=$ADMIN_PASSWORD

disable_service rabbit
enable_service qpid

HOST_IP=10.0.0.117
FLOATING_RANGE=10.0.0.224/27
FIXED_RANGE=172.16.0.0/24
NETWORK_GATEWAY=172.16.0.1
FIXED_NETWORK_SIZE=256

Also copied local.sh to devstack directory without changes - had seen the last two lines related to port 22 / ssh and felt it may be needed.

At this point I can start to play with it and hopefully make some better mistakes. I may try to get the Centos and Ubuntu Vbox VMs working.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

[hide preview]

Get to know Ask OpenStack

Resources for moderators

Question Tools

Follow
2 followers

Stats

Asked: 2014-08-08 02:51:48 -0500

Seen: 1,630 times

Last updated: Aug 15 '14