Oof ok. Thank you for helping with that Tao. I'll try to make a PR soon to ensure that the k8s tests have that set up. Is there any example in the airflow codebase I can copy from?
On Wed, Dec 19, 2018 at 1:57 AM Ash Berlin-Taylor <a...@apache.org> wrote: > NPM is needed when running off a git checkout to build the asset > pipeline. This has caused a fair amount of confusion, so I think it is > worth adding an explicit check and bailing out if the compiled files > aren't found. > > (Right now you'll see a small warning in the start up of `airflow > webserver` but this is quite easy to miss if you aren't looking for it) > > -ash > > > Daniel Imberman wrote on 18/12/2018 23:23: > > Well locally I do this (which I'm hoping to PR soon once tests are > > passing): https://github.com/apache/incubator-airflow/pull/4326. This > > allows me to connect to any k8s cluster. > > > > But for simplicity/parity with CI, you can follow these steps: > > > > export DOCKER_COMPOSE_VERSION=1.20.0 > > export SLUGIFY_USES_TEXT_UNIDECODE=yes > > export TOX_ENV=py35-backend_postgres-env_kubernetes > > KUBERNETES_VERSION=v1.10.0 PYTHON_VERSION=3 > > ./scripts/ci/kubernetes/setup_kubernetes.sh > > ./scripts/ci/kubernetes/kube/deploy.sh > > > > This will give you airflow runniung on minikube. you can then do a > kubectl > > port-forward <airflow container> 8080 8080 to port-forward into the > > instance. > > > > I can attempt to check webserver logs again when I have some time > > (hopefully this weekend/next week) as well. > > > > Thanks Tao! > > > > > > > > On Tue, Dec 18, 2018 at 3:15 PM Tao Feng <fengta...@gmail.com> wrote: > > > >> I assume so. How do you run your test? I could try it on my side and > see if > >> I could reproduce the problem. > >> > >> On Tue, Dec 18, 2018 at 2:17 PM Daniel Imberman < > danielryan2...@gmail.com> > >> wrote: > >> > >>> Do we need to install NPM? Should I add npm to the docker > image/bootstrap > >>> script? > >>> > >>> On Tue, Dec 18, 2018 at 1:20 PM Tao Feng <fengta...@gmail.com> wrote: > >>> > >>>> I just try with latest master which could show the dag after login as > >>>> Admin. One last check: have you done the `npm install` and `npm run > >>> build`? > >>>> On Tue, Dec 18, 2018 at 12:50 PM Daniel Imberman < > >>>> daniel.imber...@gmail.com> > >>>> wrote: > >>>> > >>>>> I believe I am using the new UI since I have to login as admin to get > >>>> in. I > >>>>> can still see certain admin panels, but the primary DAG one doesn't > >>> show > >>>>> any DAGs even when I explicitly put one in the folder. > >>>>> > >>>>> On Tue, Dec 18, 2018, 12:26 PM Tao Feng <fengta...@gmail.com wrote: > >>>>> > >>>>>> Are you using the Old UI or the new UI? Do you login as Admin user > >>> for > >>>>> new > >>>>>> UI? > >>>>>> > >>>>>> On Tue, Dec 18, 2018 at 11:56 AM Daniel Imberman < > >>>>>> daniel.imber...@gmail.com> > >>>>>> wrote: > >>>>>> > >>>>>>> Hello all, > >>>>>>> > >>>>>>> I've been testing the K8sExecutor and I notice that when I open > >> the > >>>> UI, > >>>>>> I'm > >>>>>>> unable to see any of the DAGs even though they are in the > >>> aiflow/dags > >>>>>>> folder. I'm also still able to launch DAGs via the API. I have > >> not > >>>>> tested > >>>>>>> this on any of the other executors yet. Is this a known issue? > >> Also > >>>>>> raises > >>>>>>> the question on whether we can set up some form of UI testing > >>> since I > >>>>>> don't > >>>>>>> think the current unit/integration tests would catch this kind of > >>>>> error. > >>>>>>> Thank you > >>>>>>> > >
