Display issue with overview pipeline stages, Linux Chrome only (version 51.0.2704.84).

/cc @sg00dwin @ALRubinger

I have a project/namespace dedicated to build and share all my custom images using the registry-viewer role to make my images "public".

Now I would like to be able to use these image on the web console from an other namespace.

But for now the web console only search in the local namespace and the "openshift" namespace.

It could be great to have a way to use images or templates from another specified namespace.

Regards,

I've seen this a few times now, recently in #33.

1)  authenticated e2e-user new project when creating a new project should be able to show the create project page
  - Error: Wait timed out after 3028ms

https://ci.openshift.redhat.com/jenkins/job/test_pull_requests_origin_web_console/8/console

We have a lot of this:

DataService
  .get('something', $routeParams.something, context)
  .then(function resolve(theThing) {
     // do lots of things...
    watches.push(DataService
      .watchObject('something', $routeParams.something, context, function(theThing) {
         // do lots of the same things again...
      }));      
  }, function reject(err) {

  });

I'm guessing this is to get at any cached data if we have it while waiting for the watch. I'm wondering if we can eliminate the duplication by one of these:

// option 1: 
// ---------------------------------
// get data out of cache (DataService._data(key)) then watch it
$scope.theThing = DataService.getCached(itsName);
DataService
  .watchObject('something', itsName, context, function(theThing) {
    $scope.theThing = theThing;
  }, reject);




// Option 2: reuse resolve/reject functions
// ---------------------------------
var resolve = function(theThing) {
  // do stuff
};
var reject = function(err) {

}

DataService
  .get('something', itsName, context)
  .then(resolve, reject);
DataService
  .watchObject('something', itsName, context, resolve, reject);




// option 3:
// ---------------------------------
// perhaps we have to ensure the GET happens before the WATCH
// so that we dont get out of sync data. 
// has an extra anonymous function but otherwise seems clean
  var resolve = function(theThing) {
    $scope.theThing = theThing;
    // do other happy stuff...
  };
  var reject = function(err) {
    // do sad stuff...
  }

  DataService
    .get('something', itsName, context)
    .then(function(theThing) {
      resolve(theThing);
      DataService
        .watchObject('something', itsName, context, resolve, reject); // reuse resolve, reject
    }, reject);

Prob option 3 would win us the most cleanup.

Thoughts @jwforres @spadgett ?

TODOs to resolve this issue

• update controllers/build.js (PR #133)
• update controllers/buildConfig.js
• update controllers/deployment.js
• update controllers/deploymentConfig.js
• update controllers/image.js
• update controllers/persistentVolumeClaims.js
• update controllers/pod.js
• update controllers/pods.js
• update controllers/route.js
• update controllers/service.js

Instructions:

Open Overview page
Expand service
Observe as shown

Google Chrome   51.0.2704.84 (Official Build) (64-bit)
Revision    9125d1d23608a4fe105afb75b0c480a8dd3eef93-refs/branch-heads/2704@{#705}
OS  Linux 

Original comment is here: https://github.com/openshift/origin-web-console/pull/82/files#r69194937

Some of our template logic may be getting out of hand. Here is one example from browse/_deployment-details.html:

<button 
  ng-if="'deploymentconfigrollbacks' | canI : 'create'" 
  ng-show="!rollBackCollapsed && (deployment | deploymentStatus) == 'Complete' && !(deployment | deploymentIsLatest : deploymentConfig) && !deployment.metadata.deletionTimestamp" 
  ng-disabled="(deploymentConfigDeploymentsInProgress[deploymentConfigName] | hashSize) > 0" 
  type="button" 
  class="btn btn-default btn-xs" ng-click="rollBackCollapsed = !rollBackCollapsed">

The ng-show in particular has a lot of details. I think we could probably reduce this down to <button ng-show="showRollBack">Roll back</button> and use a clearly named function to set the boolean in the controller.

Thoughts @jwforres @spadgett

Hi,

Is it documented somewhere how you have generated (openssl, keytool, ...) your server.crt & server.key containing the SSL Certificate ?

Usually event messages are a sentence or two, but it's possible for them to be very long. We should handle this better. For example,

/cc @jwforres

The documentation link under the help menu should be opening in a new tab but right now its opening in the current tab.

Using examples/jenkins/pipeline/pipelinetemplate.json.

/cc @jwforres

The new Command Line Tools page looks great! Nice work, @jhadvig!

I noticed that we're duplicating some information on the About page about the CLI, however.

Perhaps we revise that last sentence to:

With the OpenShift command line interface (CLI), you can create applications and manage OpenShift projects from a terminal. To get started using the CLI, visit the Command Line Tools page.

Thoughts, @jwforres, @spadgett, @benjaminapetersen, @sg00dwin?

In previous version if you clicked on a route link it opened in a new tab/window. In this version 1.3.0.alpha.2 opens on same which is annoying.

cc: @spadgett @jwforres

Original issue openshift/origin#5412

Per @spadgett comment, there is an empty state for tables in patternfly:
https://www.patternfly.org/patterns/table-view/

On the add to project page the tags on images and templates are actually links, this means the right click context menu -> open in new tab option is accessible. However the href will just point the user back to the project list page (homepage). We could:

1. Disable the right click context menu on those links
   or
2. Set the href to the current page plus a query param for a filter

I prefer we do (2) since its reasonable to want to link to the page with a tag pre-filtered (e.g. from docs / blog posts / etc)

Steps to reproduce:

1. Log into OpenShift Console
2. Create new project with some default values
3. Go back to Home
4. Delete the project you'd just created
5. Status of the project shows marked for deletion
6. Item does not disappear on its own when it's deleted; console needs a refresh

When you are on the project settings page and choose "delete" from the drop down menu, a modal appears asking you to type the project name. The input does not automatically receive focus.

As a result of our project requirements, I need the pod in the pageThe DetailsThe EnvironmentLogsThe EventsPut a label, the introduction of the company server page links, what should I do?

See openshift/origin#9425

/cc @jwforres

This happens when a new pipeline build completes and is added to the build trends chart while on the build config page. The bar does not show up, and there is no duration value in the tooltip when hovering. Refreshing the page causes the bar to appear.

?tab=environment etc. doesn't work on the browse builds page.

We either need to copy this over:

    // Check for a ?tab=<name> query param to allow linking directly to a tab.
    if ($routeParams.tab) {
      $scope.selectedTab = {};
      $scope.selectedTab[$routeParams.tab] = true;
    }

Or come up with a way to eliminate that block from all our controllers, maybe by creating a selected-tab directive that can be added to uib-tabset?

<uib-tabset selected-tab="selectedTab"> <!-- ideally a directive could check the url param and set uib-tabset... -->
  <uib-tab active="selectedTab.details" heading="baz">
    stuff...
  </uib-tab>
  <uib-tab active="" heading="bar">
  </uib-tab>
    things...
</uib-tabset>

Thinking out loud.
@rhamilto

For extension purposes, is there a better way to access variables such as user name?

Docs have lead me towards making something like this which feels a little hacky

// Fake add to get access to $scope
HawtioExtension.add('nav-help-dropdown', function($scope) {
    openshiftName = ($scope.$parent.user.fullName || $scope.$parent.user.metadata.name);
    return null;
});

Can i add more custom categories to "Add to Project" page more easily from external point instead of modifying create.js for them? If not can we have a feature of having add/delete these categories from UI directly & use them?

On the 'Other Resources' Page, if after page load the window is resized in any way (horizontal or vertical) the 'Choose a Resource' dropdown will expand to 100% of the page width when clicked. See screenshot.

Occurred on Fedora 23 in Chrome & Firefox

@jwforres @spadgett

I have done this in command line:

[vagrant@origin ~]$ oc new-app kubernetes/guestbook
--> Found Docker image 4305190 (19 months old) from Docker Hub for "kubernetes/guestbook"

    * An image stream will be created as "guestbook:latest" that will track this image
    * This image will be deployed in deployment config "guestbook"
    * Port 3000/tcp will be load balanced by service "guestbook"
      * Other containers can access this service through the hostname "guestbook"
    * WARNING: Image "kubernetes/guestbook" runs as the 'root' user which may not be permitted by your cluster administrator

--> Creating resources with label app=guestbook ...
    imagestream "guestbook" created
    deploymentconfig "guestbook" created
    service "guestbook" created
--> Success
    Run 'oc status' to view your app.

In the UI I see this:

Inmediately I do a scaling:

[vagrant@origin ~]$ oc scale kubernetes/guestbook --replicas=10
the server doesn't have a resource type "kubernetes"
[vagrant@origin ~]$ oc scale dc/guestbook --replicas=10
deploymentconfig "guestbook" scaled
[vagrant@origin ~]$ oc get dc
NAME        REVISION   REPLICAS   TRIGGERED BY
guestbook   1          10         config,image(guestbook:latest)
[vagrant@origin ~]$ oc get pods --no-headers | wc -l
1

I can see that now there¡s 10 pods requested, but still 1 pod.

UI Does not reflect this. THIS IS THE ERROR/ISSUE

In the UI I see this:

Then after 2 minutes (or so, state consolidation):
I see the console updated:

[vagrant@origin ~]$ oc get pods --no-headers | wc -l
10

If I scale now to 20 from CLI:

[vagrant@origin ~]$ oc scale dc/guestbook --replicas=20
deploymentconfig "guestbook" scaled
[vagrant@origin ~]$ oc get dc
NAME        REVISION   REPLICAS   TRIGGERED BY
guestbook   1          20         config,image(guestbook:latest)
[vagrant@origin ~]$ oc get pods --no-headers | wc -l
20

I see immediate change in console.

/cc @jwforres @fabianofranz

Ben notes that the user actually has access to templates in other projects that you have view rights on; might be nice to have a comprehensive view of all templates at the user's disposal.

Not critical.

Overview page has lost all information about the image being run, or the build.

Looking at this overview page I get no valuable information. Maybe that's intended, but I think that even with hover (like in the graphs) could be better than nothing. I need to navigate away to get some information.

Think of overview as the dashboard that I do not want to leave, unless I have an specific action to do, and there should be links to most of them (like deployments and current deployment) or pod or service, but nothing else.

version: 1.0.3.alpha.2

I have an app with 3 deployments and 1 service. When I link main deployment/service with a second, I can not add a 3rd one. I can do it editing the service by hand via cli, but not through UI.

Observed on current Origin latest Docker image (upstream/master). Unsure what corresponding version of the newly-extracted origin-web-console applies.

1. As a user who is not logged in, request a page from the console, for instance:

https://10.1.2.2:8443/console/project/default/overview/

2. Observe the console redirect to the login page:

https://10.1.2.2:8443/login?then=%2Foauth%2Fauthorize%3Fclient_id%3Dopenshift-web-console%26response_type%3Dtoken%26state%3D%257B%2522then%2522%253A%2522%252Fproject%252Fdefault%252Foverview%2522%252C%2522nonce%2522%253A%25221464575117627-188965410684100422701140277249369410143364625528371304029883494082064318640%2522%257D%26redirect_uri%3Dhttps%253A%252F%252F10.1.2.2%253A8443%252Fconsole%252Foauth

3. Enter credentials (ie. admin/admin) and hit "Log In". Observe interstitial redirect:

https://10.1.2.2:8443/console/oauth#access_token=56EXHu57Ddj0nNS3a06lGI2FshWIJicPNfdy79QtAKI&expires_in=86400&state=%7B%22then%22:%22/project/default/overview%22,%22nonce%22:%221464575217400-4009554928223396578558102190174732909820355798996169206124878501852858850331%22%7D&token_type=Bearer

4. End result is the Console Overview page:

https://10.1.2.2:8443/console/

Was expecting end result of the initially-requested page:

https://10.1.2.2:8443/console/project/default/overview/

Requesting @spadgett triage. :)

I have 2 routes, first one is long (as typical with some of our cluster naming) and can't see info about secondary routes.

The message field is now being used to show what triggered the last deployment on a DC. We need to stop checking for the message field to trigger the warning icon. Eventually this will be replaced by Conditions but currently no warnings are being set so just comment out the warnings for now until we have ones we can show.

We currently use the build config label to associate builds to their build config, but because labels only support 64 chars and build config names can be up to 256 the label may be a truncated value. The annotation openshift.io/build-config.name will have the full value. Since that annotation was only recently added we need to fallback to looking at the label if the annotation doesn't exist.

I see a blank page. Not sure if you need to delete the service, too, or only the deployment config.

To reproduce:

1. Add to project
2. Pick any builder
3. Use default source repo
4. Let the build / deployment finish. The deployment appears on the overview.
5. Delete the service
6. Delete the DC

Result: The overview page is empty.

/cc @jwforres

Hi,
I find it difficult to think that the ? icon has important things to show. I would rather have some top level items, and then maybe some in dropdown, like this console (at the bottom in this case):

• Download the client as top level, with drop down to different platforms:
• Documentation link
• Submit an issue link
• And about (where we could add release notes) and make easy to extend

The icons aren't vertically aligned when one of the stage names is long.

/cc @sg00dwin

Numbers like 30m touch the edges of the circle:

/cc @sg00dwin

Hi,

I'm new to origin-web-console. I just setup testing environment in cloud with --hostname=0.0.0.0 and --baseUrl=. However when i visit https://ip:9000/ i got following error

{"error":"invalid_request","error_description":"The request is missing a required parameter, includes an invalid parameter value, includes a parameter more than once, or is otherwise malformed.","state":"{\"then\":\"/\",\"nonce\":\"1467047314345-3673651079364708862158794790216584555681010689653270092627925064274853237928925\"}"}

the request URL is

https://ip:8443/oauth/authorize?client_id=openshift-web-console&response_type=token&state=%7B%22then%22%3A%22%2F%22%2C%22nonce%22%3A%221467047314345-3673651079364708862158794790216584555681010689653270092627925064274853237928925%22%7D&redirect_uri=https%3A%2F%2F52.196.72.38%3A9000%2Foauth

thanks!

@sg00dwin fyi, probably introduced by the fixed table layout

It would be interesting to have a way (direct link) from the overview page and to see information about current build.

The 'openshift' namespace is actually configurable in the master config OpenShiftSharedResourceNamespace

We need to be passing that configuration through the dynamic config.js and using that as the namespace we pull from, and look for anywhere else that we make assumptions about what namespace that is.

When we update the metrics chart, we request the full data for the time range (e.g. 1 hour) and redraw the chart. Hawkular will break the data into n buckets, and we show the usage for each bucket. This means the exact start and end of each bucket changes on every update, and past data can appear to change slightly since the buckets start and end times have changed.

It would be better to keep the previous data and flow new data to the chart by requesting only metrics from the end timestamp of the last bucket we have. This way previous data wouldn't change. Only new data would be added to the chart. (We can remove buckets from the front of the chart when they fall out of the time range as well.)

@jwforres @fabianofranz

Per this comment https://github.com/openshift/origin-web-console/pull/133/files#r69573887

@jwforres

@liggitt suggested moving the drag handles to the left of the field since they're right beside a destructive button.

/cc @jwforres @ajacobs21e @rhamilto @benjaminapetersen

Ideally the states would toggle seamlessly, but angular is applying an ng-animate before the ng-hide, which has the effect of both states being visible briefly.

In projects where someone is only running builds and has no DCs, no Services, and no non-infrastructure pods. You got a blank overview. I suspect this is because our check on the hashSize of pods doesnt take into account that we dont show you build and deploy pods today.

To reproduce:

• Browse to a deployment config and choose Edit YAML from the Actions menu in the upper right
• Edit an env var
• Click Save
• Click Actions and Edit YAML again
• Make an additional edit to an env var
• Click Save
• Notice the resulting error

I notice the error isn't thrown once the Deploy button is re-enabled following the deployment resulting from the first edit.

@benjaminapetersen

On the overview page, the progress bar shows animation when things are running. When you go to the build page though and look at the list of the builds, the progress bar doesn't show animation. It is an inconsistency first, but it also makes you wonder if things are actually running still.