NoNoNoo's profile - activity

2016-05-05 12:02:40 -0500 received badge  Nice Answer (source)
2016-04-05 16:31:49 -0500 received badge  Famous Question (source)
2015-09-17 20:56:14 -0500 received badge  Nice Answer (source)
2015-04-10 16:48:05 -0500 received badge  Nice Answer (source)
2015-04-02 02:23:45 -0500 answered a question why snapshot become so huge considering original image is small

You can use virt-sparsify You have to create the instance snapshot, export it to a qcow2 file, sparsify it and reimport the qcow2 file.

2015-03-26 14:20:46 -0500 answered a question Horizon Clustering Solutions

Horizon is completely stateless and can be scaled horizontally, with requests being load-balanced across some servers. Here there is an example with HA Proxy

There are different GUI solution/tools offered by different vendor that allow to install openstack cluster with HA feature for all the openstack service (nova, neutron, glance, ecc):

2015-03-13 04:46:38 -0500 answered a question What speed switch to run nova instances on NAS

1 Gbit ethernet has a bandwitch of 1 Gbit/s that is equal to 128 MB/S of bandwitch for each hypervisor.

If you confgure 10 VM on each hypervisor you have an average storage bandwitch of 12,8 MB/s for each VM not considering any overhead.

Currently the magnitude order throughput of a traditional magnetic disk is 100 MB/s. If you want use this infrastructure in production using VM with moderate IO I think you need 10 Gbit connection to shared storage device.

2015-03-13 02:15:42 -0500 commented answer Build centos 6 cluster on openstack

From centos documentation you can use iscsi target for qdisk but I never implemented it.

2015-03-12 14:43:08 -0500 answered a question Build centos 6 cluster on openstack

There are some limitation using a cluster software on two openstack instances with centos 6 image.

They depends from cluster software that you want use, with centos 6 there are three software cluster that you can use:

  • cman + service manager For two node cluster you need a quorum disk that require a shared disk between the instances that can't be implemented easily in Openstack. You should use unicast UDP for heartbeating because I'm not sure that multicast isn't supported in Openstack. There isn't any fence agent available for openstack instances.

  • corososync + pacemaker It doesn't require a quorum disk but the problem of fence agent remains.

  • keepalived you can try to follow the following guide but I think that you keepalived has limited features compared to the first two options

2015-03-05 13:24:31 -0500 answered a question how to add RHEV cluster as a compute node in my existing openstack icehouse setup

Nowadays you can't do it with openstack. As you can see RHEV is not supported as hypervisor for KVM

About the integration between openstack and rhev there is only a blueprint:

With this slideset from HP

2015-02-25 09:42:23 -0500 answered a question CentOS-6-x86_64-GenericCloud Openstack Image Credentials required

The normal method to authenticate to this images(and many other openstack images) is using keypair

An alternative is setting a password for root in the qcow2 image using the guestfish tools

2015-02-23 06:36:51 -0500 answered a question Docker and Cinder, is it possible in openstack?

I haven't used yet openstack with docker containers but I watched openstack hypervisor support matrix ( ).

For docker hypervisor the feauture "attach volume" and "detach volume" aren't supported.

2015-02-23 06:27:08 -0500 answered a question Running my own script for ceilometer alarm

Ceilometer alllows to invoke a web url when an alarm change state. You can see the ceilometer alarm creation options( ):

--alarm-action <Webhook URL>
URL to invoke when state transitions to alarm. May be used multiple times. Defaults to None.

--ok-action <Webhook URL>
URL to invoke when state transitions to OK. May be used multiple times. Defaults to None.

You can use also Heat in order to set the alarm-action and the ok-action url as you can see in Heat documentation.

You have also to develop a web application that expose an url that you can invoke from ceilometer alarm. The language used for your application depends on your programming skill on different languages. You can see some of these solutions:

  • A J2EE application server like tomcat
  • A light web server(nginx or apache) with a php script
  • A javascript web appplication wroted using Node.js
2015-02-19 06:51:27 -0500 received badge  Famous Question (source)
2015-02-12 05:02:14 -0500 answered a question No bootable device.

I list some task that can help you to troubleshoot the problem:

2015-02-11 02:31:52 -0500 answered a question Should phys ports on the control plane have the same name?

Renaming network interface name in Centos 7 requires different tasks respect to Centos 6.X

Here there is the tasks that I tested few minutes ago: ( in my example i rename the interface with mac address 00:0c:29:35:7e:1a to eth2)

  • Edit /etc/sysconfig/grub and add net.ifnames=0 biosdevname=0 in the variable GRUB_CMDLINE_LINUX
  • Execute grub2-mkconfig -o /boot/grub2/grub.cfg
  • Reboot
  • Create the file /etc/udev/rules.d/70-persistent-net.rules and add the following line SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?", ATTR{address}=="00:0c:29:35:7e:1a", ATTR{type}=="1", KERNEL=="eth", NAME="eth2"
  • Reboot
  • Rename network configuration file ifcfg-* in /etc/sysconfig/network-scripts to ifcfg-eth2
  • Edit the file /etc/sysconfig/network-scripts and rename the device name with eth2
  • Reboot
2015-02-10 05:04:08 -0500 answered a question What are different VM allocators available in Openstack?

Have you already read the documentation related to nova scheduler algorithms based on filters and weights?

Here there is some references:

If you have already read it, you can try to implement you own custom filter and weight function modifying the small examples provided in the second and third links

2015-02-10 04:32:55 -0500 answered a question how we can write an api in openstack

There is already a openstack project called Ceilometer ( ) and its main goal is collect measurements of the utilization of the physical and virtual resources of openstack infrastructures.

It has several metrics already implemented for example memory and cpu usage of a instance, network metrics of a instance and so more.

You could follow the following roadmap:

  • Install ceilometer on your infrastructure ( ) and verify which metrics are already collected by default
  • Try to verify if there are already some integration between ceilometer and LBaaS (for example I found )
  • Search documentation related to add custom metrics on ceilometer in order to avoid to "reinventing the wheel"
2015-02-06 07:47:38 -0500 answered a question Can I use openstack for self service provisioning to Physical Systems?

Ironic is the name of the openstack project used to manage bare-metal server Here there is an installation guide

Here there is the status of the project reported from the first link:

Ironic was first available as an Incubated project in the "Icehouse" release. The following "Juno" release saw improved integration with Nova, as well as many stability and feature improvements. Ironic is officially Integrated with OpenStack beginning with the "Kilo" release.

Icehouse, 2014.1: Juno, 2014.2:

Information on the status of Ironic's test suite can be found here:

For a list of current drivers and the status of their CI testing, see the wiki page we maintain and the Stackalytics page which tracks third-party CI

In my experience I haven't see yet Ironic implemented in any commercial distribution of Openstack. It is used by HP Helion but only for compute node deploy.

2015-02-06 07:24:17 -0500 answered a question Dashboard overview

This limits also called in openstack quota are defined for each openstack project.

You can modify it using command line or using the web interface

2015-02-05 01:58:16 -0500 answered a question Openstack juno's on centOS 6.5

The previous version of openstack, Icehouse, is supported by Centos 6.X. The minimum requirement for installation of Juno on a Centos/Redhat is Centos/Redhat 7.0.

There is some exception to this for example Mirantis Openstack 6.0 ( ) provides a Centos 6.5 with Openstack Juno but I think that Mirantis uses a custom rpm repository.

2015-02-04 04:56:54 -0500 received badge  Scholar (source)
2015-02-04 04:56:48 -0500 received badge  Notable Question (source)
2015-02-04 04:56:48 -0500 received badge  Famous Question (source)
2015-02-04 04:56:48 -0500 received badge  Popular Question (source)
2015-02-03 09:41:45 -0500 commented answer Openstack Instance High Availability(How to make pets VM highly available?)

Yes, i think to it but I didn't found any opensource/commercial/draft solution to implement it. I think that could be useful for enterprise use of openstack. If you find more detail about this solution please share it with me. Thanks.

2015-02-03 09:36:39 -0500 answered a question Is Active/Active & Active / Passive form of migration is possible in openstack without downtime ?

Nowadays there isn't any openstack component that automatically detect failed compute nodes and execute the migration (or VM evacuation ) of the VM executed in the failed host.

There is the command "nova evacuate" that require shared storage but it's a command that an admin should execute for each vm.

For automatic evacuation there are only openstack blueprint ( ) or custom solution ( ).

In order to obtain high available openstack instances, you can use vCenter as Openstack compute node ( ) and configura a vSphere cluster with two or more ESXi with HA feature enabled, but there is some limitation about using it with openstack.

2015-02-03 09:14:08 -0500 answered a question Openstack Instance High Availability(How to make pets VM highly available?)

I think that there isn't yet the best solution for VM-High availability in openstack. You have listed some of solution but they should be checked in each context to verify if they are valid.

  • About "nova evacuate" command that isn't a automatic way of instance fail-over you can develop a custom software in order to verify compute nodes failure(example: nagios, a monitoring software) and automatically execute "nova evacuate".I have read somenthing about this solution here
  • For an HA-mysql you can try to setup a maria db galera cluster with ha proxy and keepalived, here there is a tutorial for centos 6 ; i haven't tried it on openstack.
2015-02-02 10:50:52 -0500 commented answer Fuel deployment: host unreachable on br-mgmt

Yes if you use VLAN tag in fuel you have to configure also these vlan on the switch. Are you using fuel installation of mirantis openstack? You can "untag" br-mgmt and br-storage but you need other 2 interfaces(for example eth2 and eth3) for each openstack server

2015-02-02 10:47:24 -0500 received badge  Citizen Patrol (source)
2015-02-02 09:52:50 -0500 answered a question Fuel deployment: host unreachable on br-mgmt

The ports of the managed switch connected to eth0 interfaces should be configured in trunk mode in order to allow the following types of vlan traffic:

  • br-mgmt with vlan tag 1199
  • br-storage with vlan tag 1198
  • br-fw-admin that is traffic sended without any tag so you have to set any pvid for these ports(you wrote the pvid 1011 I think)

Using HP/Comware syntax you have to configure each switch port connected to eth0 interfaces as in the following example:

[HP]interface FuelEth0InterfaceName
[FuelEth0InterfaceName]port link-type trunk       # Configure trunk lynk type 
[FuelEth0InterfaceName]undo port trunk permit vlan 1  # Remove default vlan 1
[FuelEth0InterfaceName]port trunk permit vlan 1199    # Allow tagged traffic for vlan 1199
[FuelEth0InterfaceName]port trunk permit vlan 1198    # Allow tagged traffic for vlan 1199
[FuelEth0InterfaceName]port trunk pvid vlan 1011      # Allow untagged packets and assign to it the vlan id 1011
[FuelEth0InterfaceName]undo shutdown

You maybe have eth0 switch ports configured in access mode that allow only the packets without vlan tag, indeed you can ping two fuels nodes using br-fw-admin interfaces. You should configure these ports on the switch in trunk mode.

2015-01-31 22:14:46 -0500 received badge  Notable Question (source)
2015-01-30 05:44:34 -0500 received badge  Popular Question (source)
2015-01-29 09:49:45 -0500 asked a question Redhat subscription support on other vendor openstack distribution

I'm searching information related to the support of Redhat operating system as instance(virtual machine) on other vendor openstack distribution.

I already know that Redhat distributes the software "Redhat Openstack Platform" and offers commercial support for it but I'm searching an alternative to Redhat.

I'm specially interested in Mirantis Openstack distribution that use the operating system centos 6.5 for controller and compute nodes (or vCenter and ESXi in the case of VMware compute nodes). But in general I'm interested to know if there is some opestack distribution supported by RedHat.

I know that Redhat distribute a qcow2 image and it can run without problems in several other openstack distribution but I'm interested in support in the meaning of paying the subscription, register the operating system subscrition and open tickets if occurs some problem.

I have seen also that Redhat offers two type of subscription ( ):

  • Red Hat Enterprise Linux Server (2 sockets, 1 physical or 2 virtual nodes)
  • Red Hat Enterprise Linux for Virtual Datacenters (2 sockets, unlimited virtual, no physical )

I would know if I can use this second subscription type that is more useful for installation with several openstack instances. I have seen in Redhat site that this license is supported for VMWare, Hyper-V and RHEL KVM Support. Can I use it with VMWare compute nodes or centos KVM compute nodes?

Have you been some experience on this argument?

2015-01-29 07:16:59 -0500 answered a question Make flat network available only to some tenants/users

From openstack documentation :

Currently the concept of 'external' network is somewhat similar to the concept of a 'shared' network. However, while every tenant can operate on a shared network, performing operations such as creating port, the set of operations a tenant can perform on an external network is more limited, as it's currently restrained to setting external gateways on routers and creating floating IPs.

Nevertheless, the concept of 'external' implies some forms of sharing, and this has some bearing on the topologies that can be achieved. For instance it is not possible at the moment have an external network which is reserved to a specific tenant. That external network will always show up in queries performed by other tenants too.

2015-01-29 04:03:56 -0500 received badge  Famous Question (source)
2015-01-28 08:45:31 -0500 asked a question Scheduling instance snapshot backup to an external repository

I'm searching some software (commercial, opensource or only some draft) that allow to backup openstack instances using snapshot. The openstack command line allow to create snapshot and export them to a qcow2 file then you can backup it using a backup software or copy the qcow2 file to an external storage. I'm searching a solution to automatize and to schedule this process.

I think that I can do it creating some custom bash script and some existing tool(example a backup software or something like rsync) but I would know if someone has arleady worked on something similar in order to reinvent the wheel.

2015-01-28 01:46:16 -0500 answered a question Keepalived on openstack vm is possible?

Other people have been your problem . It seems caused by the compute node firewall that block vrrp protocol used by keealived. You can use the iptables rules described in the link to enable vrrp protocol in each compute node but I don't know if vrrp protocol are forwarded between different compute node.

In alternative you can use other cluster software for instance corosync/pacemaker that can use unicast udp for the heartbeat connection.

2015-01-26 09:49:47 -0500 answered a question I wanted to configure CSR1000V on openstack cloud. Can somebody post me a link to configure CSR1000V on openstack

Have you read the official documentation ?

It reports the list of commands to install it on Openstack.

2015-01-26 07:02:38 -0500 commented question Horizon run very slowly

Try to post the output of top of your 3 nodes in or other similar website and share the output here. To extract the output of top and sava it in a file you can use the command top -n 1 > top-output.txt

2015-01-25 04:39:19 -0500 received badge  Notable Question (source)
2015-01-23 19:07:22 -0500 received badge  Student (source)
2015-01-23 18:55:14 -0500 received badge  Popular Question (source)
2015-01-23 09:12:55 -0500 commented question ping -c 4 controller. This command is unable to ping

I think that the OS is Ubuntu watching the content of /etc/hosts. Have you configured the ip address as ip address of one of your network interface? Have you tried to ping ?

2015-01-23 03:14:25 -0500 commented answer KVM installed in blade server but not available in virtual box VM's

QEMU should be slower than KVM in general but I think that it's ok for learning. KVM or QEMU are used by openstack only for execute the operating system of the instances. All the other component of openstack(horizon,neutron, network topology page and so on) shouldn't be affected by KVM o QEMU choice

2015-01-23 02:10:48 -0500 answered a question KVM installed in blade server but not available in virtual box VM's

You need to extend virtualization feature of your physical cpu to your virtualbox virtual cpu in order to use KVM Hypervisor inside a virtual machine of another hypervisor (virtual box in your case); this feature is called also nested virtualization.

From this url I have seen that virtual box doesn't have nested virtualization feature.

You have two other options:

1) Use another hypervisor that support virtualization(VMware Workstation support it)

2) Use QEMU instead of KVM for Openstack that doesn't require virtualization cpu features

2015-01-21 06:19:58 -0500 edited question Red Hat Enterprise Linux OpenStack Platform 5 reference architecture or tutorial for HA installation

I have downloaded the trial version of "Red Hat Enterprise Linux OpenStack Platform 5" in order to study a solution for an enterprise high-available openstack environment with RHEL instances.

If I have understood correctly, the unique supported method to install an high-available opestack environment with RedHat Software is using "Red Hat Enterprise Linux OpenStack Platform Installer" based on foreman. Is it correct?

I have tried to launch several times the automated openstack installation in HA mode using "Red Hat Enterprise Linux OpenStack Platform Installer" but it always fails with different error during controller nodes installation.

I don't know if the problem is the limited resources of virtual machines where I run the installation or a wrong configuration of the architecture. Unfortunately I can't open a ticket to the Redhat support because I'm using a trial version for learning.

Do you have any tutorial, reference architecture, configuration description, blog article related to Openstack High-available installation using "Red Hat Enterprise Linux OpenStack Platform 5"?

I have already read the official documentation ( ).

I searched already using google or youtube but I found really few technical webpages related to RHOSP 5 and nothing related to the High Availability instalation. I'm interested also in understand how Redhat Enterprise Openstack Platform is used in enterprise openstack installation compared with other vendors( for example mirantis, rackspace, hp helion).