[Openstack] METADATA 500 ERROR (neutron dhcp related)

yang sheng forsaks.30 at gmail.com
Fri Mar 24 15:56:51 UTC 2017


I have a Opentack (kilo) running in multi-server environment for a while
and everything are working fine.

I am using flat network with ML2 plugin only.

configured HA of DHCP agents in neutron.conf:


Basically, 2 dhcp agents each with a unique ip in that subnet in my simple
network, VM can get the ip from dhcp access METADATA service .

[root at controller001 /]# neutron dhcp-agent-list-hosting-net simple-network
| id                                   | host                  |
admin_state_up | alive |
| 158ad39f-e36c-4ca5-828d-b51381f74a40 | controller001 | True           |
:-)   |
| 6120e511-84f6-4bd2-8d46-df3e14abbfa6 | controller002 | True           |
:-)   |

2 dhcp agents` IPs are: ip_1, ip_2


Inside VM, the static route to is one of the 2 dhcp agent's
ip ( I guess the one provide dhcp lease offer). If that DHCP agent goes
down for any reason or the DHCP agent's IP changed,  I am not able to curl
the metadata service, and it returns 500 Internal server error.

I have to manually bring up the dhcp-agent on that controller. More
importantly, the ip of  dhcp agent's can't be changed because of the static
route in many VMs.

I can create an automation script to check the dhcp agents and it's IP.
However, is there better way to keep persistency of the dhcp agent status
and ip?

If using L3 agent with OVS, will this issue get solved?

Thanks for any advice.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20170324/57ee3db3/attachment.html>

More information about the Openstack mailing list