# Revision history [back]

Just a comment
Quoting starts

Here I’d like to summarize our six-months long experience with the TripleO installer:

    In our experience, the configuration of the OpenStack cluster using only the TripleO installer is not flexible enough. As a workaround, we ended up writing a bunch of Ansible scripts.
The overcloud update can take a very long time to complete and it can fail because of random errors. Also during the update operation all the overcloud nodes have to be reachable by the TripleO installer. For this reason, I personally cannot imagine using TripleO to manage a cluster with more than one hundred nodes.
As we experienced, the TripleO installer can easily break a working OpenStack cluster. This is a big no-no for a production system.

Overall, I think that the TripleO installer in the Mitaka version of OpenStack would need more work to become production ready. In the meantime, we’re continuing with patching of what we have.


In the future, there are other projects that could replace the TripleO installer. I found the kolla-ansible and kolla-kubernetes rather promising.

Quoting ends

I guess that major concern of any customer is safe production maintenance and abilty recover from disaster for sure with minimal downtime.
I am also very sorry about RH's policy regarding packstack development. Ales Nosek mentioned "Kolla" related projects and there is also well known project running by Google && Intel && Mirantis "implementation Openstack on Kubernetes containers"

Just a comment
Quoting starts

Here I’d like to summarize our six-months long experience with the TripleO installer:

  In our experience, the configuration of the OpenStack cluster using only the TripleO installer is not flexible enough. As a workaround, we ended up writing a bunch of Ansible scripts.
The overcloud update can take a very long time to complete and it can fail because of random errors. Also during the update operation all the overcloud nodes have to be reachable by the TripleO installer. For this reason, I personally cannot imagine using TripleO to manage a cluster with more than one hundred nodes.
As we experienced, the TripleO installer can easily break a working OpenStack cluster. This is a big no-no for a production system.

Overall, I think that the TripleO installer in the Mitaka version of OpenStack would need more work to become production ready. In the meantime, we’re continuing with patching of what we have.


In the future, there are other projects that could replace the TripleO installer. I found the kolla-ansible and kolla-kubernetes rather promising.

promising.

Quoting ends

I guess that major concern of any customer is safe production maintenance and abilty recover from disaster for sure with minimal downtime.
I am also very sorry about RH's policy regarding packstack development. Ales Nosek mentioned "Kolla" related projects and there is also well known project running by Google && Intel && Mirantis "implementation Openstack on to Kubernetes containers" containers"

Just a comment
Quoting starts

Here I’d like to summarize our six-months long experience with the TripleO installer:

        In our experience, the configuration of the OpenStack cluster using only the TripleO installer is not flexible enough. As a workaround, we ended up writing a bunch of Ansible scripts.
The overcloud update can take a very long time to complete and it can fail because of random errors. Also during the update operation all the overcloud nodes have to be reachable by the TripleO installer. For this reason, I personally cannot imagine using TripleO to manage a cluster with more than one hundred nodes.
As we experienced, the TripleO installer can easily break a working OpenStack cluster. This is a big no-no for a production system.

Overall, I think that the TripleO installer in the Mitaka version of OpenStack would need more work to become production ready. In the meantime, we’re continuing with patching of what we have.
In the future, there are other projects that could replace the TripleO installer. I found the kolla-ansible and kolla-kubernetes rather promising.


Quoting ends

I guess that major concern of any customer is safe production maintenance and abilty recover from disaster for sure with minimal downtime.
I am also very sorry about RH's policy regarding packstack development. Ales Nosek mentioned "Kolla" related projects and there is also well known project running by Google && Intel && Mirantis "implementation Openstack on to top Kubernetes containers"