VNC and firewalld on Centos 7 compute node

By default firewalld service is enabled and may cause troubles in accessing via VNC to instances running on Centos compute node.


By default zone=public is used, so it is enough just to add 5900/5901 ports or/and vnc-server service to this zone:


Nova installation on Centos 7


After starting rabbitmq-server service it is not possible to change password using:


In file /etc/systemd/system/ change

to – add sudo

And restart rabbitmq-server:


Nova incompatibility

Packages for nova which are downloaded with yum contain nova version which is incomaptible with that is running on controller node (Ubuntu):

Controller (Ubuntu 14.04):

Compute (Centos 7):

Problem is described here:

Install packages from EPEL repo manually (only 3 days older):

Packages to install:

I had to forcefully install them using rpm -ivh –force then I restarted openstack-nova-compute service.
After this instances spawned correctly, yet I could not reach them (iptables rules missing, so I disabled iptables).

DHCP addresses got distributed to the instances correctly and I could reach internet from them via my Ubuntu-AIO which acts as a gateway to the internet as well.

Trove installation on Ubuntu 14.04

When installing Trove service on Ubuntu 14.04 I encountered a problem with dependencies not met between Trove related packages. 

Thus the following packages were installed manually with dpkg -i <package_name>

In order to make apt-get still working after manual package installation (which does not check dependencies) it is a must failed dependecies for manually installed packages

In /var/lib/dpkg/status, for each package for which there is a missing dependency just remove it (lsb-base (>= 4.1+Debian11ubuntu7)) from line in the status file:

If there is any config file missing in /etc/trove then download it from:



In order to use RabbitMQ the following line must be added in

  • trove.conf
  • trove-conductor.conf
  • trove-taskmanager.conf

Instead of (as per manual):

Swift AIO

OpenStack installation instructions from assume 3 node architecture therefore and the biggest problems that can be encountered while performing AIO (all-in-one) installation are with Swift.

In order to perform AIO installation for Swift the following instructions must be followed:

With some modifications according to normal installation instructions of Juno on Ubuntu (just use common sense to modify e.g. IPs accordingly, add keystone authentication as it is missing in AIO). After this installation command swift stat works properly.

If any swift related process does not start then it needs to be ensured that e.g. log directories/files have owner set to swift:swift.

Use Swift as backend for Glance

Use Swift as backend is described in the following post:

Additionally in swift store must be added in /etc/glance/glance-api.conf:

Lastly, comment out any existing swift related configuration in glance-api.conf

Heat installation on Ubuntu 14.04

First of all as rpc_backend option in heat.conf file must be set as follows:

Another problem is the same as with Nova in post:  i.e. heat-engine service does not start unless mysql service starts first. This is the same problem as with nova-scheduler and nova-cert.


Heat-engine is a SysV service thus we need to first check mysql priority (in the example below it is 19):

 In all appropriate runlevels (2,3,4,5) add a symbollic link /etc/rcX.d directories so that heat-engine start *AFTER* mysql service:

 or use:

 Even though heat-engine process seems to be UP:

Service status is shown as STOP/WAITING:

Due to some reason heat-engine service is called by some different service during bootup twice before mysql is up and it fails to launch (even though in boot log it shows [OK]



After mysql is launched again heat-engine is started (thanks to adding it /etc/rcX.d with appropriate priority):

 All in all service works fine but its status is shown incorrectly.

Instance not started due to OVS database problem

The following problem may appear while attempting to start an instance:


Just restart ovs service:

DHCP not working on demo-net

DHCP-agent works in a separate namespace but is connected to br-int in ovs, after creating a demo-net it can be seen that an interface on which dhcp (tap3b4d7bf0-12) server works is assigned to VLAN 4095 (so called “dead-vlan”) on the br-int making it inaccessible for a running instance which has an interface in demo-net.

Changing tag statically to 1 (same tag as used by qr* interface where DHCP requests from an example instance are seen with tcpdump):

Does not solve the problem, DHCP requests reach DHCP agent and are replied but replies are not forwarded to the instance (probably some more static changes are required in OVS)

DHCP port (in this example is visible in neutron as DOWN:


Rather simple one. Disable and enable back DHCP service on the network (I did it from horizon) -> after this DHCP port’s status changes to ACTIVE (and probably something more internally in OVS is changed so as to make DHCP replies being forwarded properly to the instance).

Instance not started due to Neutron misconfiguration on Ubuntu 14.04

While attempting to start instance the following problem appears:




Add in /etc/nova/nova.conf in [DEFAULT] section:

Problem described here: