[Openstack] openstack mitaka high memory issue

Clint Byrum clint at fewbar.com
Mon Aug 22 13:30:09 UTC 2016

Excerpts from Jose Manuel Ferrer Mosteiro's message of 2016-08-22 08:43:57 +0200:
> In my case I choose Mitaka because Ubuntu 16.04 LTS has 5 years of free
> support without updates breaking anything. I can program updates and
> reboots, go home, receive zabbix messages at 18:00 saying the services
> had been rebooted and are OK, ... and sleep well. 
> I tried the automatic "yum update" with CentOS but they changed some
> version of some neutron related package and some config file var.
> Networking began to fail while I was at home. :-( 
> In my case the version is not important because people does not use it
> as a real cloud. They use it to deploy virtual machines, like a remote
> Virtualbox with balancing as a service to deploy OpenShift and
> Kubernetes. 
> Maybe I am the only person that uses heat to deploy. :-( 

Heat is certainly popular, though not as popular as the core services.

I wonder, could you try setting this in your heat config:

convergence_engine = true

This will switch Heat to using the experimental convergence engine,
rather than the legacy one. One of the main goals of convergence is to
make Heat lighter weight. It has also received most of the attention
over the past few cycles, and I wonder if the refactoring done on Heat
to support it has not taken into account something in the legacy flow.

Anyway, if that does fix the problem, perhaps report that as a bug here:


Might create a compelling case for switching it on by default in Newton.
It's not too late! :)

> The question is, your system will be used as a virtual machines deployer
> like a classic vmware deployment or as an IAAS like AWS or GCE? If it
> will be used like a tipical vmware deployment in my opinion any
> OpenStack supported version could be the right choice. 

Why does it have to be an "or" question? Are they really that different?

More information about the Openstack mailing list