On 07/10/2017 04:31 PM, Monty Taylor wrote:
On 07/10/2017 01:21 PM, Ed Leafe wrote:
On Jul 10, 2017, at 5:06 AM, Mikhail Fedosin <[email protected]
<mailto:[email protected]>> wrote:
Given all the advantages and features of Glare, I believe that it can
become the successful drop-in replacement.
Can you clarify this? Let’s assume I have a decent-sized deployment
running Glance. If I were to remove Glance and replace it with Glare,
are you saying that nothing would break? Operators, users, scripts,
SDKs, etc., would all work unchanged?
I also have this question. The glance API is one of the most fundamental
and basic APIs. You pretty much can't do anything useful on a cloud
without touching it.
Actually - as an easy first-step - set up a gate job with a devstack
that has glare and no glance and run shade's functional tests against
it. We're pretty darned lenient - if you can pass our functional tests
then talking about stricter things like tempest is worthwhile. If you
can't - hopefully there will be some clear areas to work on.
That said - it's not like glare couldn't also do those things - but I'd
need to understand some real specifics about what a cloud switching from
glance to glare looks like to the end user.
Also, we have a new upload API designed for glance that took a LARGE
amount of wrangling to get consensus on. I'd also want to know what this
situation looks like in glare, if image upload in glare supports all of
the use-cases that we figured out image upload in glance needed to
support. AND - there are folks who want import-from which was removed
between glance v1 and v2. Does glare support something in this area?
__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: [email protected]?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
