Ask Your Question
0

add compute node

asked 2016-04-28 01:44:13 -0600

luckyyd gravatar image

Hi All,

I deploy the rdo openstack on one server (say controller0 and compute0). And all the components worked fine. The VMs are working fine,too.

And when I tried to add a compute node (say compute1), I followed the RDO "Adding a compute node".

The packstack script also showed that "Installation completed successfully" without error

However, I can not ping the VMs that in compute0, it seems the router in compute0 goes wrong.

If I shut down em2 in controller0, all things worked. If I start em2 for compute1, VMs are failed.

And if I created an instance in compute1, vnc login is also failed.

The ip configuration of controller0:

br-ex: inet 10.156.64.53
em1: N/A
em2: inet 172.16.0.10

compute1:

em1: inet 10.156.64.70(DHCP)
em2: inet 172.16.0.20

And the configuration of packstack file:

CONFIG_NOVA_COMPUTE_PRIVIF=em2
CONFIG_NOVA_NETWORK_PRIVIF=em2
CONFIG_CONTROLLER_HOST=10.156.64.53
CONFIG_COMPUTE_HOSTS=172.16.0.20
CONFIG_NETWORK_HOSTS=10.156.64.53
EXCLUDE_SERVERS=10.156.64.53

I do not know how to debug this issue. Any help please?

edit retag flag offensive close merge delete
add a comment

1 answer

Sort by ยป oldest newest most voted
0

answered 2016-05-23 12:32:09 -0600

dbaxps gravatar image

Fist of all
Your statement "I deploy the rdo openstack on one server (say controller0 and compute0)" is confusing
I would guess you did AIO RDO Mitaka setup and attempted to add Compute Node
As far as br-ex belongs 10.156.64.0/24 ( I guess /24 ) then this is external network
So no DHCP for 10.156.64.0/24
172.16.0.0/24 ( I guess /24 ) is mgmt network
Adding server block

EXCLUDE_SERVERS=172.16.0.10
CONFIG_CONTROLLER_HOST=172.16.0.10
# Adding 172.16.0.20 
CONFIG_COMPUTE_HOSTS=172.16.0.10,172.16.0.20   
CONFIG_NETWORK_HOSTS=172.16.0.10

RDO Mitaka already contained patch https://review.openstack.org/#/c/257033/ , which is bringing back to life EXCLUDE_SERVERS directive, presumes original deployment of ML2&OVS&VXLAN type
Following ( or similar block ) should be present in answer-file

CONFIG_NEUTRON_ML2_TYPE_DRIVERS=vxlan
CONFIG_NEUTRON_ML2_TENANT_NETWORK_TYPES=vxlan
CONFIG_NEUTRON_ML2_MECHANISM_DRIVERS=openvswitch
CONFIG_NEUTRON_ML2_FLAT_NETWORKS=*
CONFIG_NEUTRON_ML2_VLAN_RANGES=
CONFIG_NEUTRON_ML2_TUNNEL_ID_RANGES=1001:2000
CONFIG_NEUTRON_ML2_VXLAN_GROUP=239.1.1.2
CONFIG_NEUTRON_ML2_VNI_RANGES=1001:2000
CONFIG_NEUTRON_L2_AGENT=openvswitch
CONFIG_NEUTRON_LB_INTERFACE_MAPPINGS=
CONFIG_NEUTRON_OVS_BRIDGE_MAPPINGS=physnet1:br-ex
CONFIG_NEUTRON_OVS_BRIDGE_IFACES=
CONFIG_NEUTRON_OVS_TUNNEL_IF=em2
# Following entry is required by mentioned patch 
CONFIG_NEUTRON_OVS_TUNNEL_SUBNETS= 172.16.0.0/24 
CONFIG_NEUTRON_OVS_VXLAN_UDP_PORT=4789

So I assume that VTEPs subnet is the same as MGMT network ( one more guess )

edit flag offensive delete link more
add a comment

Your Answer

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

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

subscribe to rss feed

Stats

Asked: 2016-04-28 01:44:13 -0600

Seen: 272 times

Last updated: May 23 '16

Related questions

vm connection error between vms on different hosts (vxlan tunnels)

RDO packstack very slow write on cinder volumes

RDO allinone

RDO Installation problem on CentOS 7 because of MongoDB

No vnc on second compute node

can not boot cirros

RDO allinone (havana) with neutron

RDO Error: php56w-common conflicts with php-common-5.4.16-36.el7_1.x86_64

Compute Instance not found when trying to launch trove db instance on Icehouse release

install openstack juno version with RDO fail