If we stick with the —mobile flag proposal setting up the service broker and default org is not enough. Don’t we need to install our custom resource definitions too?Am 12.01.2018 um 13:21 schrieb Matthias Wessendorf <mwessend@redhat.com>:I was about to say, that I'd be not surprised if there is not that much interest in brining this in...Currently they have "oc cluster ACTION [options]",and for the "up" action, they list a bunch of options, including installing (experimental) bits, like: Cassandra/Hawkular), via the --metrics optionOn Fri, Jan 12, 2018 at 1:15 PM, Craig Brookes <cbrookes@redhat.com> wrote:While that is an interesting proposal around the addons, I would suggest waiting for the OpenShift team to get back on your issue. Adding a plugin structure to oc is likely a much bigger piece of work and need a lot more engagement from the OpenShift team.That doesn't mean a different proposal for this approach could not be made however, but rather that our current proposal covers our immediate need and has clear prior art.On Fri, Jan 12, 2018 at 12:06 PM, Matthias Wessendorf <mwessend@redhat.com> wrote:I'd actually prefer more --mobile - which than does exactly what we need--On Fri, Jan 12, 2018 at 1:04 PM, Phil Brookes <pbrookes@redhat.com> wrote:I’ve discussed this privately with Wojciech as I had misunderstood his suggestion, so here’s my attempt to clarify the discussion:
Rather than inserting our mobile oriented tasks into the upstream tool (oc), which doesn’t provide value to all users of that tool but rather only to the users who are interested in the mobile use-case; we should instead modify the tool so that it can accept addons, similar (ideally, identical) to minishift addons.
The proposal here would be something like:
oc cluster addons install mobilecore oc cluster addons enable mobilecore oc cluster upThere a few benefits to this solution:
- There is a clear value-add to the upstream tool
- We only need to maintain one mobilecore plugin for both minishift and oc cluster up
- The barrier to entry for mobile developers is low
- The process for setting up a local cluster is the same with both tools
- Future changes to the setup process of the Mobile-Core can be submitted to our own repo
@Wojciech please correct me, if I’ve still misunderstood anything!
Regards,
Phil.On Fri, Jan 12, 2018 at 11:55 AM, Craig Brookes <cbrookes@redhat.com> wrote:+1On Fri, Jan 12, 2018 at 11:46 AM, David Martin <davmarti@redhat.com> wrote:I think there's great value in --mobile flag.Its the same approach for metrics and logging (--metrics, --logging)It would do everything needed to enable the mobile extension and get the cluster into the state we want.Having to provide multiple flags would make it just that little more complex that it raises the barrier.I do think there's upstream value in allowing configuring a default org for the broker though, but I don't think it should be part of this proposal.--On 12 January 2018 at 11:39, Phil Brookes <pbrookes@redhat.com> wrote:Hey Wojciech,
Thanks a lot for the feedback.
If I have understood you correctly, you are suggesting splitting this task in to two discrete pieces of work:
- Add parameters to oc cluster up, to be able to configure the ansible-service-broker with custom docker credentials and docker org to pull in the ASBs from.
- Add the —mobile flag to enable the Mobile-Core UI extension.
With the intention of making this an easier change to adopt.
If that is correct then I am happy to proceed this way and I can update the proposal to reflect this as a route to implementation.
Maybe it would be best to bring this discussion to the proposal so that the OpenShift team also have access to it?
Thanks,
Phil.On Fri, Jan 12, 2018 at 11:21 AM, Wojciech Trocki <wtrocki@redhat.com> wrote:Mobile flag as it doesn't provide ability to change values and imposes some hard dependencies.That may be hard accept and maintain on upstream.
Upstream works by generalization so if we want to contribute there it's best to pass something that can be used by many people.How about `oc cluster up --service-broker=true --service-broker-org=aerogearcatalog` and whatever else we need? That's easier to accept and contribute upstream.I have done changes for cluster up recently so if you need anything in that let me know - we can work on this together.
Another idea is to specify oc cluster up pre/post actions (hooks) that will allow people to run ansible roles for things we need.
That will be solid proposal that upstream may consider.
Developers can still have simple command by doing alias:alias oc-mobile-up=`oc cluster up --service-broker=true --service-broker-org=aerogearcatalog`
On Fri, Jan 12, 2018 at 10:51 AM, David Ffrench <dffrench@redhat.com> wrote:Fantastic Phil, This is 100% needed in my opinion. The mobile services need to be a first class citizen in OpenShift to ensure easy setup and adoption.DAVID FFRENCHsenior software engineer, RED HAT MOBILE
Communications House, Cork RoadWaterford, Ireland
Hey Everyone,
I have opened a proposal with the OpenShift team, hoping to start a discussion around the addition of a
--mobile=trueflag to theoc cluster upcommand. Runningoc cluster up --mobile=truewill start a local OpenShift cluster with all of the MCP features enabled.This would be a great way for us to gain some exposure for the Mobile Core and get it into the hands of developers more quickly and easily, so I’d love to see you share your thoughts on this idea. You can review / comment / support it here:
https://github.com/openshift/origin/issues/18089 Thanks,
Phil.Project lead AeroGear.org
_______________________________________________
feedhenry-dev mailing list
feedhenry-dev@redhat.com
https://www.redhat.com/mailman/listinfo/feedhenry-dev
--David MartinRed Hat MobileTwitter: @irldavemIRC: @irldavem (#aerogear)--Craig BrookesRHMAP@maleck13 GithubProject lead AeroGear.org--Craig BrookesRHMAP@maleck13 Github--Project lead AeroGear.org
_______________________________________________
feedhenry-dev mailing list
feedhenry-dev@redhat.com
https://www.redhat.com/mailman/listinfo/feedhenry-dev
