← Back to team overview

openstack team mailing list archive

Re: Bringing High Availability to OpenStack Keystone and Glance

 

On 04/09/2012 05:02 AM, Martin Gerhard Loschwitz wrote:
Am 04.04.12 09:28, schrieb Alex Glikson:
We have a prototype of similar capabilities targeting Nova services, using ZooKeeper (heartbeat/membership, leader election, etc -- will share with the community soon). Would be interesting to compare the two approaches.

Regards,
Alex



Hello Alex,

great to hear that! Comparing the two approaches will definetely be an
interesting thing once we have comparable solutions for both methods
available (i.e. glance or keystone integration for ZooKeeper or Nova in-
tegration for Pacemaker).

The reason I started with Glance and Keystone is the fact that I expect
it to be particularly more difficult to catch all odds and ends when it
comes to integrating Nova into Pacemaker. For example, verifying that
a process called "nova-network" is running isn't difficult, but making
sure that on the very same system exactly the network-configuration is
in place (iptables-wise etc. pp.) that is necessary to consider
nova-network running is a lot harder. I'm completely open if anyone has
ideas on how to fiddle with this the best way :)

Best regards
Martin

For either approach, would running the Services in HTTPD make it easier, harder, or no different to maintain HA?





References