----- Original Message -----
On St 6. srpen 2014, 16:13:34 CEST, Radek Vokál wrote:
> Vasku, I like the idea and the workflow. If I understand the workflow
> right - the layered image will build on top of base image picked in COPR
> (eg. Fedora20 or CentOS7)? Also would it make sense to run our own docker
> registry and have in connected to docker.io?
My idea was to "guess" base image automatically from the
os-version(-arch) selected by user in COPR - so if the copr build is
for fedora-20-x86_64, we can say yes, it's 64bit, and use fedora:20 as
If we really want to build images, it would make sense to serve them in
private registry. On the other hand, I've changed my mind since I sent
this proposal - I am not sure if Fedora infrastructure can/will handle
building and storing possibly hundreds or thousands of images. I'd like
to start with a feature that will prepare and let you download a
Dockerfile which you can then modify locally and build your image.
Well, not everyone will request to build an image and correct me if I'm wrong, but for
updates docker only stores diffs.
I need to talk about this with Mirek and people from Fedora
> ----- Original Message -----
>> Could you please have a look at attached diagram briefly describing
>> potential architecture of the image building service? Would this be a
>> viable solution?
>> On 24.7.2014 10:42, Václav Pavlín wrote:
>>> On St 23. červenec 2014, 13:33:06 CEST, Richard Marko wrote:
>>>> On 07/23/2014 10:44 AM, Nick Coghlan wrote:
>>>>> On 07/23/2014 01:42 AM, Václav Pavlín wrote:
>>>>>> What do you think? Do you think this makes sense for copr?
>>>>> If we could use COPR to make Beaker packages for Fedora and EPEL
>>>>> upload Dockerfiles to generate Docker images for the various
>>>>> that would be seriously cool. On the other hand, it might make more
>>>>> sense as a separate service that reacts to fedmsg events indicating
>>>>> COPR builds are available.
>>>> +1 for separate service.
>>> I am not against separate service. The separate service would probably
>>> simply mean running Docker in some cloud instance and call it's API
>>> through docker-python bindings and then collect the results.
>>> Lead Infrastructure Engineer
>>> Developer Experience
>>> Brno, Czech Republic
>> Lead Infrastructure Engineer
>> Developer Experience
>> Brno, Czech Republic
>> copr-devel mailing list
> copr-devel mailing list
Lead Infrastructure Engineer
Brno, Czech Republic