Ask Your Question

unable to ssh to an instance (seems to hang)

asked 2014-09-22 12:51:52 -0500

mickael gravatar image

Hello all,

I am very new to openstack ... and I am trying to install icehouse in a 3-nodes model (1 controller, 1 network, 1 compute). The 3 nodes are implemented as VM on a single host.
The host is running fedora 20 + KVM & libvirt (with nested virtualization enabled).
The 3 cloudstack nodes are running fedora 20 basic install.
3 virtual networks are defined with libvirt as follows:
- "default" network --> providing NAT to the host (CIDR is
- "management" network --> network dedicated for management activities (CIDR is
- "data" network --> network dedicated for cloudstack VMs communications and access to 'outside' world though the network node (CIDR is

The controller node is connected to the "default" & "management" networks
The network node is connected to the "default", "management" & "data" networks
The compute node is connected to the "management" & "data" networks

All installation steps are done and all components seem to run properly (openstack-status reports that all services are ok).
Then I updated the default security-group to allow all ingress & egress traffic for protocols ICMP, TCP, UDP ... Yes I know this is not that secure ;-) but just for testing purposes for now.

As the admin user, I have defined an external network (, a new tenant (tenantA) as well as a user being a member of tenantA.

As the new user (member of tenantA), I have defined a private network & subnet ( as well as a router which is connected to the private network and the external one.
I have also downloaded and registered into glance, the PoC CirrOS image and a fedora20 one. And I have generated and registered an ssh keypair (RSA)
Then, using the openstack dashboard, I spawned 2 VMs: one out of CirrOS image (private IP is and a second one out of the fedora-20 image (private IP is

When connected to the network node (using ssh),:
- I can see 2 net namespaces (one qrouter and one qdhcp)
- from the qrouter namespace as well as from the qdhcp one, I can successfully ping the private interface of the tenantA's virtual router (, the gateway to the public network (, my home internet "box" gateway ( and the public google DNS server (
- from the qdhcp namespace, I can also ping the 2 VMs that I started ( &
- lastly, still from the qdhcp namespace, when I issue a telnet on port 22 to the 2 VMs I get a connection. Here is what I get:

[root@network ssh]# ip netns exec qdhcp-d66ccdd6-c0e1-48e0-a6b4-c90e1e816fde telnet 22
Connected to
Escape character is '^]'.

My problem:
from the qdhcp namespace, when I try to ssh to both VMs, the connection seems ...


edit retag flag offensive close merge delete


try adding -vvv to your ssh to get more information

T u l gravatar imageT u l ( 2014-09-22 14:39:45 -0500 )edit

2 answers

Sort by ยป oldest newest most voted

answered 2014-09-23 01:04:02 -0500

mickael gravatar image

Thanks for your time & your comment.
Actually, I just found (this post).
Applying the solution proposed in the first answer (adjusting the MTU size) fixed my problem also on the fedora-20 instance.

Here is a quote of the solution:

create file /etc/neutron/dnsmasq-neutron.conf
edit /etc/neutron/dhcp_agent.ini:
[DEFAULT] dnsmasq_config_file = /etc/neutron/dnsmasq-neutron.conf

Then restart neutron-dhcp-agent and the VM ... For me that worked :)

edit flag offensive delete link more


Good that worked and for sharing in this forum.

SGPJ gravatar imageSGPJ ( 2014-09-27 08:13:16 -0500 )edit

answered 2015-08-12 04:18:37 -0500

sankalp82 gravatar image

Thanks, its work for me !!!

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

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2014-09-22 12:51:52 -0500

Seen: 1,548 times

Last updated: Sep 23 '14