Example config:

apiVersion: apps/v1
kind: Deployment 
metadata:
  name: sidecar
spec:
  selector:
    matchLabels:
      app: sidecar
  replicas: 2
  template:
    metadata:
      labels:
        app: sidecar
    spec:
      volumes:
      - name: shared-logs 
        emptyDir: {}
    
      containers:
      - name: app-container
        image: alpine
        command: ["/bin/sh"]
        args: ["-c", "while true; do date >> /var/log/app.txt; sleep 5;done"]
    
        volumeMounts:
        - name: shared-logs
          mountPath: /var/log
    
      - name: sidecar-container
        image: nginx:1.16.0
        ports:
          - containerPort: 80
    
        volumeMounts:
        - name: shared-logs
          mountPath: /usr/share/nginx/html

@bryanl commented on Tue Apr 16 2019

There should be some indication to the user that the port forward connection has been lost.

@mdaverde commented on Mon May 13 2019

This can be part of overall contribution doc, but I wanted an issue to capture UI specifics. We should mention:

• Getting set up
• Overview of design
• Testing requirements of changes/new features

@bryanl commented on Fri Apr 19 2019

When sugarloaf is released, enable it to be run as a kubectl plugin.

• investigate running dash as a plugin (should only require a binary name kubectl-sugarloaf
• figure out how krew works and create a plan for managing octant with krew
• Update documentation for install octant with krew and running with kubectl

There is a cohort of objects that Octant doesn't support (e.g. apps/v1 ControllerRevision). If they show up in the resource viewer, there is no way to link to them since there is no dashboard view for them. There needs to be a way to handle this case. There are a few options in this case.

1. Add these objects to the dashboard. Then they can be linked. The issue here is they might not be useful on their own. DaemonSet and ReplicaSet use ControllerRevision internally.
2. Don't visit these objects.

/cc @wwitzel3

@bryanl commented on Tue Mar 26 2019

Provide viewers for well known annotations. This could make viewing annotations more friendly to the user

Potential annotations:

• kubectl.kubernetes.io/last-applied-configuration: This will always be JSON. The frontend will have a highlighter, so there could be a hint to display this as formatted JSON. Also, the backend could send this text formatted with indenting and line breaks.

@mdaverde commented on Thu Apr 18 2019

Should the backend send a new component for each annotation type (i.e. JSON)?

@bryanl commented on Thu Apr 18 2019

Ultimately, the backend would send a component. What that component is is TBD. First you have to gather list of common annotations that could exist, then you would need to determine if our current set of components could be combined to build a better view. I'd assume we'd probably get some new components out of this, but would default to using existing components first.

@bryanl commented on Mon May 27 2019

A slow control plane can result in timeouts. Currently we log a message to the console, but don't do anything else. The solution should inform notifying the user that the control plane is timing out. Also, the dashboard should be able to provide a better experience when these timeouts are occuring and respond in a faster manner.

@wwitzel3 commented on Tue Jun 11 2019

Here is what happens now ...

2019-06-11T10:58:05.705-0400    ERROR   objectstore/watch.go:124        check access failed: client resource: unable to retrieve rest mapping: Get https://internal.kubernetes.dev-usw2.hq.7infra.com:6443/api?timeout=32s: dial tcp 10.11.81.85:6443: i/o timeout
github.com/heptio/developer-dash/internal/objectstore.(*Watch).List
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/objectstore/watch.go:124
github.com/heptio/developer-dash/internal/describer.LoadObjects
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/describer/describer.go:71
github.com/heptio/developer-dash/internal/describer.(*ObjectLoaderFactory).LoadObjects
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/describer/describer.go:41
github.com/heptio/developer-dash/internal/describer.(*List).Describe
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/describer/list.go:54
github.com/heptio/developer-dash/internal/describer.(*Resource).Describe
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/describer/resource.go:42
github.com/heptio/developer-dash/internal/describer.(*Section).Describe
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/describer/section.go:30
github.com/heptio/developer-dash/internal/describer.(*Section).Describe
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/describer/section.go:30
github.com/heptio/developer-dash/internal/modules/overview.(*realGenerator).Generate
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/modules/overview/generator.go:90
github.com/heptio/developer-dash/internal/modules/overview.(*Overview).Content
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/modules/overview/overview.go:184
github.com/heptio/developer-dash/internal/api.(*contentEventGenerator).Generate
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/api/stream.go:97
github.com/heptio/developer-dash/internal/api.(*contentStreamer).content.func2
        /home/wwitzel3/go/src/github.com/heptio/developer-dash/internal/api/stream.go:240

uses kubeContext , should use kubeConfig

@bryanl commented on Mon Feb 18 2019

Introduce config maps and secrets to resource viewer by updating the visitor to check pds for references to config maps and secrets.

@bryanl commented on Fri Mar 22 2019

Since printers are starting to form a shape, it should be possible to refactor all the printers
and reduce them to a configuration.

There are two types of printers:

• list of objects
• objects

list of objects

List of objects comprises of a table with column and rows. This may be able to stay the same as
there isn't much overlap in the different list printers

objects

Previously, the object printer was introduced. You can supply config, summary, item sections,
and enable events, pod template, or job template. It feels like you should be able to describe
printing an object with a struct and then have a single printer.

https://github.com/vmware/octant/search?q=octant-dashboard&unscoped_q=octant-dashboard

Initi containers in pods are labeled as such. Deployment should work the same way.

@GuessWhoSamFoo commented on Wed Dec 05 2018

For cronjob summaries, it would be nice to have a descriptor for schedule (e.g. */1 * * * * - every minute) or show the next scheduled time to run.

There are several places tracking directories from mockgen (.gitignore, .dockerignore, Makefile). Makefile should be able to get a list of them and have documentation on places that needs to update if those paths change.

Feature Request

Didn't see this listed anywhere and probably a nit on my part. When switching namespaces while looking at a specific resource I would expect to stay on that resource but change the namespace or have the option to configure that somewhere.

What Happened

With octant running and accessing Service Accounts in the default namespace changed namespaces to kube-system and was brought to the kube-system namespace overview

What I expected

With octant running and accessing Service Accounts in the default namespace change namespaces to kube-system and be brought to Service Accounts in the kube-system namespace

@shomron commented on Tue Oct 23 2018

It would be useful to be able to navigate from a ConfigMap / Secrets view to the workloads (Deployments/Pods/etc) that reference them. If specific keys are referenced, navigating through the keys to a subset of the above could also be useful.

@GuessWhoSamFoo commented on Fri Apr 12 2019

Although succeeded pods may be skipped in overview, they should be linked from job especially for logging purposes.

In the pod summary view, show the node it runs on.

@mdaverde commented:

There's a long list of CRDs on initial load of the dashboard, but then a significant amount of them disappear after a few seconds. This is happening to me consistently when I start Octant.

@bryanl commented on Mon May 27 2019

Polling is hard due to pile ups. Instead of polling, the frontend should register its current context (e.g. viewing a list of lists, viewing a list, viewing an object), and send changes only when there is an update.

This would require additional changes:

• timestamps should autoincrement in the front end
• possibly would need to be able to stop the current context. i.e. tell the backend to stop sending updates

@bryanl commented on Wed Jun 05 2019

Instead of polling, why not send updates when things change rather than always.

In some cases, the resources viewer doesn't link all objects. I am currently investigating why this happens.

@bryanl commented on Mon May 27 2019

Initial load time is too slow. In the browser it can take upwards of 30 seconds to display the dashboard.

@wwitzel3 commented on Wed Jun 05 2019

thought, decouple the loading of the initial objects from the generation of the connection graph

Surface the status of Octant, starting with logging information to the user in the UI.

@xstevens commented on Mon May 13 2019

This morning I had shared some thoughts in Slack and we wanted to document those here.

It would be nice from a developer's standpoint to be able to easily craft all the resources needed for an app with the UI's assistance. In this regard it might have similar goals as AWS console.

The killer feature though would be to then export the YAML configuration for the designed resources, so it could be repeatable using infrastructure-as-code processes.

This could really apply to anything, but one example might be to assist designing network policies of what services/pods need to talk to each other. Then the users could start from a highly locked down start state and easily convey what their intentions are using a drag-n-drop UI between nodes. This has been done in tools like Visio going a long way back but again was never exportable into any useful repeatable form outside of the tool itself.

@xstevens commented on Fri May 17 2019

Saw this gif demo on Konstellate's repo. This is pretty much what I was thinking of for exporting.

Create helpers that ease the creation of navigation for plugins that are also modules. Plugin authors should not have to care about the internal details of our routing to enable navigation in their module plugin.

@bryanl commented on Tue Apr 16 2019

Error message looks like:

2019-04-16T08:12:55.021-0400    ERROR   api/stream.go:255       event generator error: adding tab: unable to retrieve the complete list of server APIs: metrics.k8s.io/v1beta1: the server is currently unable to handle the request
github.com/heptio/developer-dash/internal/api.(*contentStreamer).content.func2
        /Users/bryan/go/src/github.com/heptio/developer-dash/internal/api/stream.go:255

It should be possible to handle this error in a more graceful fashion

@wwitzel3 commented on Wed Jun 05 2019

in the code that checks if a tab has an error, instead of skipping, add a component.Error to the tab.

@wwitzel3 commented on Wed Jun 05 2019

related #794

@wwitzel3 commented on Wed Jun 05 2019

maybe have a list of errors in Config with a unread counter

@bryanl commented on Wed Oct 03 2018

• Cron job
• daemon set
• deployment
• Job
• Replica set
• Replication controller
• Stateful set
• Ingress
• Service
• Config map
• PVC
• Secret
• Role
• Role binding
• Event?

When an error is encountered listing and reading CRDs we can check the status of the CRD and surface the conditions along side the error message in the status view.

related #72

@GuessWhoSamFoo commented on Fri Jun 28 2019

When running octant against a namespace on a DECC cluster, there is a case where the user does not have access to the default namespace. This results in a lengthy error output for each group.

2019-06-28T10:45:49.937-0700    ERROR   objectstore/watch.go:176        check access failed: denied {Namespace:default Group:rbac.authorization.k8s.io Resource:rolebindings Verb:list}
github.com/vmware/octant/internal/objectstore.(*Watch).List
        /home/sfoo/go/src/github.com/vmware/octant/internal/objectstore/watch.go:176
github.com/vmware/octant/internal/describer.LoadObjects
        /home/sfoo/go/src/github.com/vmware/octant/internal/describer/describer.go:75
github.com/vmware/octant/internal/describer.(*ObjectLoaderFactory).LoadObjects
        /home/sfoo/go/src/github.com/vmware/octant/internal/describer/describer.go:45
github.com/vmware/octant/internal/describer.(*List).Describe
        /home/sfoo/go/src/github.com/vmware/octant/internal/describer/list.go:74
github.com/vmware/octant/internal/describer.(*Resource).Describe
        /home/sfoo/go/src/github.com/vmware/octant/internal/describer/resource.go:50
github.com/vmware/octant/internal/describer.(*Section).Describe
        /home/sfoo/go/src/github.com/vmware/octant/internal/describer/section.go:35
github.com/vmware/octant/internal/describer.(*Section).Describe
        /home/sfoo/go/src/github.com/vmware/octant/internal/describer/section.go:35
github.com/vmware/octant/internal/modules/overview.(*realGenerator).Generate
        /home/sfoo/go/src/github.com/vmware/octant/internal/modules/overview/generator.go:95
github.com/vmware/octant/internal/modules/overview.(*Overview).Content
        /home/sfoo/go/src/github.com/vmware/octant/internal/modules/overview/overview.go:220
github.com/vmware/octant/internal/event.(*ContentGenerator).Event
        /home/sfoo/go/src/github.com/vmware/octant/internal/event/content_generator.go:50
github.com/vmware/octant/internal/event.runGenerator
        /home/sfoo/go/src/github.com/vmware/octant/internal/event/generator.go:86
github.com/vmware/octant/internal/event.Stream.func1
        /home/sfoo/go/src/github.com/vmware/octant/internal/event/generator.go:54

After switching namespace on UI, octant updates and removes default from dropdown.

Octant may need to check for available namespaces before checking access with object store

@bryanl commented on Fri May 31 2019

@bryanl commented on Sat Sep 22 2018

The octant client should be able to check if it is outdated, and update itself to the current version.

@bryanl commented on Mon May 27 2019

Sporadically, the dashboard becomes unresponsive when switching namespaces. Need to dig in the causes why and figure out how to make it quicker.

@bryanl commented on Tue Apr 16 2019

Currently only the granted permissions are shown

Xref: Homebrew/homebrew-core#42319

• 30 watches
• 30 forks
• 75 stars
• Update readme

Enable a dark theme option.

@bryanl commented on Thu May 02 2019

Include:

• [] requests
• [] limits

@mdaverde commented on Fri May 03 2019

Can we add labels/tags to these new issues? It's easier to organize the Zenhub board and see which areas need attention.

@bryanl commented on Fri May 03 2019

Yes. That would be the next step. This was created in new issues because it isn't yet triaged.

Octant 0.4 does not load for me:

Console is full of these errors:

2019-07-23T17:54:24.950+0200	ERROR	event/generator.go:103	event generator error	{"err": "get CRD from object store: cannot convert int64 to float64", "errVerbose": "cannot convert int64 to float64\nget CRD from object store\ngithub.com/vmware/octant/internal/describer.CustomResourceDefinition\n\t/go/src/github.com/vmware/octant/internal/describer/customresource.go:61\ngithub.com/vmware/octant/internal/describer.(*crdList).Describe\n\t/go/src/github.com/vmware/octant/internal/describer/crd_list.go:57\ngithub.com/vmware/octant/internal/describer.(*CRDSection).Describe\n\t/go/src/github.com/vmware/octant/internal/describer/crd_section.go:62\ngithub.com/vmware/octant/internal/describer.(*Section).Describe\n\t/go/src/github.com/vmware/octant/internal/describer/section.go:35\ngithub.com/vmware/octant/internal/modules/overview.(*realGenerator).Generate\n\t/go/src/github.com/vmware/octant/internal/modules/overview/generator.go:95\ngithub.com/vmware/octant/internal/modules/overview.(*Overview).Content\n\t/go/src/github.com/vmware/octant/internal/modules/overview/overview.go:225\ngithub.com/vmware/octant/internal/event.(*ContentGenerator).generateContent\n\t/go/src/github.com/vmware/octant/internal/event/content_generator.go:75\ngithub.com/vmware/octant/internal/event.(*ContentGenerator).Event\n\t/go/src/github.com/vmware/octant/internal/event/content_generator.go:71\ngithub.com/vmware/octant/internal/event.runGenerator\n\t/go/src/github.com/vmware/octant/internal/event/generator.go:89\ngithub.com/vmware/octant/internal/event.Stream.func1\n\t/go/src/github.com/vmware/octant/internal/event/generator.go:54\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1337"}
github.com/vmware/octant/internal/event.runGenerator
	/go/src/github.com/vmware/octant/internal/event/generator.go:103
github.com/vmware/octant/internal/event.Stream.func1
	/go/src/github.com/vmware/octant/internal/event/generator.go:54
2019-07-23T17:54:27.792+0200	ERROR	event/generator.go:103	event generator error	{"err": "generate entries for Custom Resources: retrieving CRD names: crd conversion failed: cannot convert int64 to float64", "errVerbose": "cannot convert int64 to float64\ncrd conversion failed\ngithub.com/vmware/octant/pkg/navigation.CustomResourceDefinitionNames\n\t/go/src/github.com/vmware/octant/pkg/navigation/navigation.go:126\ngithub.com/vmware/octant/pkg/navigation.CRDEntries\n\t/go/src/github.com/vmware/octant/pkg/navigation/navigation.go:73\ngithub.com/vmware/octant/internal/octant.(*NavigationFactory).genNode\n\t/go/src/github.com/vmware/octant/internal/octant/factory.go:115\ngithub.com/vmware/octant/internal/octant.(*NavigationFactory).Generate.func1\n\t/go/src/github.com/vmware/octant/internal/octant/factory.go:83\ngithub.com/vmware/octant/vendor/golang.org/x/sync/errgroup.(*Group).Go.func1\n\t/go/src/github.com/vmware/octant/vendor/golang.org/x/sync/errgroup/errgroup.go:57\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1337\nretrieving CRD names\ngithub.com/vmware/octant/pkg/navigation.CRDEntries\n\t/go/src/github.com/vmware/octant/pkg/navigation/navigation.go:75\ngithub.com/vmware/octant/internal/octant.(*NavigationFactory).genNode\n\t/go/src/github.com/vmware/octant/internal/octant/factory.go:115\ngithub.com/vmware/octant/internal/octant.(*NavigationFactory).Generate.func1\n\t/go/src/github.com/vmware/octant/internal/octant/factory.go:83\ngithub.com/vmware/octant/vendor/golang.org/x/sync/errgroup.(*Group).Go.func1\n\t/go/src/github.com/vmware/octant/vendor/golang.org/x/sync/errgroup/errgroup.go:57\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1337\ngenerate entries for Custom Resources\ngithub.com/vmware/octant/internal/octant.(*NavigationFactory).Generate.func1\n\t/go/src/github.com/vmware/octant/internal/octant/factory.go:85\ngithub.com/vmware/octant/vendor/golang.org/x/sync/errgroup.(*Group).Go.func1\n\t/go/src/github.com/vmware/octant/vendor/golang.org/x/sync/errgroup/errgroup.go:57\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1337"}
github.com/vmware/octant/internal/event.runGenerator
	/go/src/github.com/vmware/octant/internal/event/generator.go:103
github.com/vmware/octant/internal/event.Stream.func1

@bryanl commented on Thu May 30 2019

Currently we are using the summary view as the default. Would it be possible to make the resource viewer the default in a configurable way?

Allow a user to add a context from the frontend.

Currently plugins have to return specific return types from each handler. Ideally a plugin handler would only need to return a single type of *Response.

@mdaverde commented on Mon May 13 2019

The unit tests and integration tests are fulfilling the role of testing the internal logic of the Octant application.

The plan for this issue is to write up an E2E test that can smoke test the major features of Octant. Passing this in our CI/CD process tells us that we can standup the application and a user can at least use top level functionality.

Create an http handler compatible interface that allows for registering routes and content handlers for a module plugin.

Currently you must parse the content path and do all of your own dispatching inside of your Content handler method.

Maybe able to do at the same time as #23

When a pod is deleted, shutdown the port forwards

@bryanl commented on Wed Nov 21 2018

Handling invalid kubeconfig files:

• don't panic/exit on invalid kubeconfig files
• ensure log message generated includes the fact that there is an issue with the kubeconfig file
• send invalid kubeconfig status to frontend

@bryanl commented on Mon May 27 2019

To aid with debugging, it would be helpful to show the load the dashboard is creating on the backend. It should be possible to show watches, what's being watched, and informer status.

@bryanl commented on Wed Apr 17 2019

@GuessWhoSamFoo commented on Tue May 14 2019

Current behavior:

After duplicating the plugin stub with different text, it registers twice to show two identical tabs.

Expected behavior:

Installing a new plugin should not result in duplicates of existing plugins.

@bryanl commented on Wed Jun 05 2019

One way to disambiguate is to watch for this in the plugin manager. If you see duplicated tabs, preface them with the plugin name.

@bryanl commented on Wed Jun 19 2019

Also, removing the big label for this. It isn't a big since it is working as intended. Tab names should be more descriptive than the demo’s plugin.

@bryanl commented on Wed Jun 19 2019

There is another way to approach this as well. Use a tooltip to specify which plugin the ran came from. Keep in mind that tabs use vertical space.

@mdaverde commented on Thu Jul 11 2019

When I click on a my CRD from the CRD list view, I see this in the terminal console:

E0711 16:11:23.428224   89205 reflector.go:134] github.com/vmware/octant/third_party/k8s.io/client-go/dynamic/dynamicinformer/informer.go:89: Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x4e1, Code:0x2, Cause:error(nil)} when reading response body, may be caused by closed connection. Please retry.
E0711 16:11:28.471659   89205 reflector.go:134] github.com/vmware/octant/third_party/k8s.io/client-go/dynamic/dynamicinformer/informer.go:89: Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x539, Code:0x2, Cause:error(nil)} when reading response body, may be caused by closed connection. Please retry.
E0711 16:12:35.546751   89205 reflector.go:134] github.com/vmware/octant/third_party/k8s.io/client-go/dynamic/dynamicinformer/informer.go:89: Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x605, Code:0x2, Cause:error(nil)} when reading response body, may be caused by closed connection. Please retry.

Nothing in the UI console. This error causes the UI to hang, never going to the specific detail CRD page.

@bryanl commented on Mon Jul 22 2019

This isn't a CRD issue. It's a cluster connectivity issue. Notice the log is from klog. Octant will retry a few times, but will quit after five times.

@mdaverde commented on Mon Jul 15 2019

The Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x4e1, Code:0x2, Cause:error(nil)} part makes it seem it's an internal issue. The messaging here should change.

@bryanl commented on Mon Jul 15 2019

This is a client go reported error. All we could potentially do is turn down the logging verbosity.

@mdaverde commented on Mon Jul 15 2019

If this error is a cluster connectivity error, then I wouldn't propagate this client-go error to the Octant user and instead say something that means "This is a cluster connectivity error. Here are the action steps to debug/verify/fix..."

@bryanl commented on Mon Jul 22 2019

Yes, this is true. My original comment was that this isn't a CRD issue. I'll remove that label.

@mdaverde commented on Mon Jul 22 2019

I'm not sure this is a cluster connectivity issue, given that I can run kubectl describe crd/{{crd name}} with a quick response time in comparison to the UI.

To emphasize: the UI never resolves.

@mdaverde commented on Mon Jul 22 2019

Observation: looks to be each CRD detail view results in theses errors being shown:

E0718 15:11:41.729079   78466 reflector.go:134] github.com/vmware/octant/third_party/k8s.io/client-go/dynamic/dynamicinformer/informer.go:89: Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x4db, Code:0x2, Cause:error(nil)} when reading response body, may be caused by closed connection. Please retry.
E0718 15:12:26.813312   78466 reflector.go:134] github.com/vmware/octant/third_party/k8s.io/client-go/dynamic/dynamicinformer/informer.go:89: Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x461, Code:0x2, Cause:error(nil)} when reading response body, may be caused by closed connection. Please retry.
E0718 15:12:29.497056   78466 reflector.go:134] github.com/vmware/octant/third_party/k8s.io/client-go/dynamic/dynamicinformer/informer.go:89: Failed to list *unstructured.Unstructured: Stream error http2.StreamError{StreamID:0x4d3, Code:0x2, Cause:error(nil)} when reading response body, may be caused by closed connection. Please retry.

Remove the Watch store and only have ObjectStore. This will help improve overall performance and memory usage.

@mdaverde commented on Thu May 02 2019

This is in reference to the events coming in from the EventSource stream:

objectDeleted fits the warning we give to the user: Kubernetes object was deleted from the cluster.

It also seems like the server sends a redirect path. We should consider just bumping the user up to a valid parent navigational item.