OpenStack RDO stability [closed]

asked 2014-01-27 17:47:23 -0500

cloudssky gravatar image

updated 2014-01-29 05:15:24 -0500

Hello everyone,

This is my first post here, I'm playing with OpenStack since Diablo release and I'm still not convinced to use OpenStack in production and recommend it to my customers, but I'm using the RDO Havana release for a Dev environment with many troubles regarding stability and maintanance and I would like to know:

If someone is running OpenStack RDO in production using VLAN or GRE?

If not, is someone running RHOS in production?

If yes, what is the difference between RDO packstack and RHOS installation (different puppet and foreman modules)?

If there is any difference, why?

Do I need RHEL and pay for RHOS to have a stable OpenStack setup?

Is RDO packstack designed only for playing around OpenStack?

Question continued on Tue Jan 28:

Many thanks for the clarification below regarding RDO and RHELOSP. I can guess that the Foreman Installer works on RHEL and somehow the KVM version on RHEL might have extended features and is more stable as on CentOS ;-)

So regarding my Dev & Test environment, I'm using RDO on top of CentOS 6.5 on 4 nodes (1 contoller, 3 compute) with a Juniper switch using VLAN. Each servers has 3 Nics.

To come back to my problem, what happens currently on this small environment is, that sometimes either MySQL or Keystone dies, or after 2 - 3 days one of the hosts becomes unresponsive and I've to reboot the controller or the compute machines, a restart of all services doesn't often help. And the most annoying thing is, that if a VM runs into an error state a reboot doesn't help and resetting the state doesn't too and to get rid of the VM is a special art of its own. But today the whole thing works and I'm waiting for the next crash :-)

But to give some positive aspects which I appreciante using RDO is:

The deplyoment is very easy and you can get a 4 node deplyoment up and running in less than 2 - 3 hours. You can reuse the answer file for other deployments easily. Neutron seems to be very stable.

What I couldn't cover till now and couldn't find any good documentation about it was how to extend an exsitng RDO installation with an additional node using PackStack (without having to deal with IPtables and update the DB by hand :-)).

The other thing which I was very disapointed about it, was the fact that an upgrade form Grizzly to Havana didn't work and I know the upgrade is still a big issue by OpenStack and I'm very curious if this will change by Icehouse.

Thanks! Arash

edit retag flag offensive reopen merge delete

Closed for the following reason the question is answered, right answer was accepted by koolhead17
close date 2014-01-29 05:22:34.668337

Comments

Which platform are you using RDO on, Fedora or CentOS/Scientific Linux/RHEL?

sgordon gravatar imagesgordon ( 2014-01-28 08:41:25 -0500 )edit

This question is not really a question. Unless it's edited to be less of a flamebait I'd suggest to close it to avoid useless debates.

smaffulli gravatar imagesmaffulli ( 2014-01-28 16:43:24 -0500 )edit

O.K. the question has been edited.

cloudssky gravatar imagecloudssky ( 2014-01-29 04:19:11 -0500 )edit

2 answers

Sort by ยป oldest newest most voted
2

answered 2014-01-28 09:16:39 -0500

sgordon gravatar image

updated 2014-01-28 13:09:40 -0500

larsks gravatar image

Disclaimer: I work for Red Hat.

I can't answer your questions about VLANs and GRE support as they are non-specific in nature, so it's unclear whether you are hitting issues with the underlying platform, the RDO packaging, or OpenStack itself. I would suggest asking about these separately and providing specifics (logs, error messages, observed performance, etc.) so that others may offer targeted assistance.

There are users running both RDO and Red Hat Enterprise Linux OpenStack Platform (RHELOSP) in production environments. Both RDO and RHELOSP provide PackStack for proof-of-concept (PoC) deployment and Foreman for more involved production deployments, the differences between the RDO and RHELOSP versions of these deployment tools on a functional level are kept to a minimum.

The main differences between the offerings are:

  • RHELOSP provides an avenue for formal support including backporting of patches in response to customer requests, over and above those included in the standard upstream stable updates (upstream stable updates are provided to both RDO and RHELOSP users).
  • RHELOSP offers a longer support lifecycle - 1.5 years for RHELOSP 4 (Havana) [1].
  • RHELOSP offers a stack of formally certified hardware, software, and services - further extending your formal support options [2].
  • RHELOSP deploys on RHEL, RDO deploys on both Fedora and EL-derived systems (the fast moving nature of Fedora does however cause some issues for RDO users from time to time).

TL;DR RDO is community supported, RHELOSP is formally supported by Red Hat - though many Red Hat engineers work on both, just as many Red Hat engineers work on upstream OpenStack [3]. Refer to the RDO FAQ for more information [4].

  • [1] https://access.redhat.com/site/support/policy/updates/openstack/platform/
  • [2] http://marketplace.redhat.com/openstack/
  • [3] http://stackalytics.com/?release=icehouse&metric=commits&project_type=openstack&module=&company=red+hat&user_id=
  • [4] http://openstack.redhat.com/Frequently_Asked_Questions
edit flag offensive delete link more
0

answered 2014-01-29 05:22:10 -0500

koolhead17 gravatar image

I am closing this thread @sgordon has provided all the relevant information.

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

Stats

Asked: 2014-01-27 17:47:23 -0500

Seen: 914 times

Last updated: Jan 29 '14