[Openstack] Multiple availability zones

Turbo Fredriksson turbo at bayour.com
Thu Sep 29 13:45:25 UTC 2016


On Sep 29, 2016, at 2:11 PM, Tobias Urdin wrote:

> Yes because a availability zone does not "belong" to anything.


Well, the comments in the config file(s) seem to differ from that
statement:

	# Default value of availability zone hints. The availability zone
	# aware schedulers use this when the resources availability_zone_hints
	# is empty. Multiple availability zones can be specified by a comma
	# separated string. This value can be empty. In this case, even if
	# availability_zone_hints for a resource is empty, availability zone
	# is considered for high availability while scheduling the resource.
	# (list value)
	default_availability_zones = nova

	# Availability zone of this node (string value)
	availability_zone = nova

Or at least, might be a little .. "fuzzy". Does the "Multiple az can be specified
with a comma" belong to the 'default_availability_zones' (the suffixing 's' in
that seem to indicate that - plural) or to the 'availability_zone'??

How can there be multiple defaults (!!) but only one specified!?


So what you're saying is that if I specify, for example:

	default_availability_zones = nova,users
	availability_zone = nova

then that service (Neutron in this case) should be able to deal with both my
AZs? What about the i/o from Neutron? Will it be automatic, or do I have to
'link' my two Neutron (etc) controllers somehow? Or are they linked "via"
RabbitMQ and/or MySQL?
-- 
Turbo Fredriksson
turbo at bayour.com





More information about the Openstack mailing list