dashboard should read from the metrics APIs in order to be compatible with future kubernetes versions.
heapster is being deprecated (https://github.com/kubernetes/heapster/pull/2022) in favor of metrics-server (https://github.com/kubernetes-incubator/metrics-server).
The current heapster or metrics-server releases can provide the metrics APIs (https://github.com/kubernetes/community/blob/master/contributors/design-proposals/instrumentation/resource-metrics-api.md). So a single client implementation can support both on I think kubernetes 1.7 or above.
cc #2976, #1310
I can start working on this.
Awesome, this is yours then. Enjoy :) Talk to me or @konryd on slack if you need any help.
@ibrasho looking forward to your good news.
I hadn't had much time to work on this in the past couple of weeks. I'm expecting to be able to put more time into it soon.
I'd appreciate someone with good understanding of the current implementation to run ideas by on Slack or email.
+1
+1
+1 from me
On Aug 3, 2018, at 8:52 PM, Chris Beach notifications@github.com wrote:
+1
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.
+1
+1
+1
+1
👍
People, you don't need to each individually comment "+1" to follow this issue. Just subscribe to it using the button to the right.
@llamahunter
Or just use the thumbs up reaction to the original issue post to bump it up...
Since this didn't make it into the latest release, is it still planned to replace the dependency on Heapster with the metrics API? (@ibrasho ?)
We're having to run Heapster along-side metrics-server in our 1.11 cluster just for the dashboard :grimacing:
Any news on that ? As right now using k8 1.11.3 and the latest version of the dashboard/heapster/influxdb
I am unable to get the metric system to work.
Metric client health check failed: the server is currently unable to handle the request (get services heapster). Retrying in 30 seconds.
I just want to add that those additional dependencies take quite a bit of resource, too. I hope you will be able to switch to the new API soon. Thanks for working on this.
Cheers!
@ibrasho Do you need some help about that ?
@XciD He has not beed active since then. Perhaps you can take it with our assistance?
/cc @floreks
the default heapster install on top of a Kubeadm cluster is broken with certificate errors
this one is a must. waiting for it
Is it possible to run heapster for dashboard in parallel with metric-server? Will they give the same stats?
Any good news about this issue? Can we now able to do the same?
@XciD Were you picking this up and running with it? I haven't investigated the changes necessary but I'm happy to help. I wanted to know if you were close to pushing something though before I started making my own changes.
@johnSchnake I will not be able to start soon, so go ahead.
@johnSchnake @XciD in the light of Heapster retiring in 1.13 timeframe is there any latest progress on getting dashboard to metricsAPI? Is there an ETA of when this can land. 1.13 is targeted for release next week 12/3
I’ve just started peeking at this so I don’t have a great timeline; I do plan on actively working on it though.
On Nov 26, 2018, at 3:38 PM, Aish Sundar notifications@github.com wrote:
@johnSchnake @XciD in the light of Heapster retiring in 1.13 timeframe is there any latest progress on getting dashboard to metricsAPI? Is there an ETA of when this can land. 1.13 is targeted for release next week 12/3
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the thread.
@johnSchnake Great! Contact us on #sig-ui Slack channel for any info needed. I would recommend starting with the backend only as frontend is undergoing migration and it makes no sense to adjust anything at the moment in the current master branch.
Hey @johnSchnake - hope you have a good holiday. Do you have any update on this work?
/assign
This is actively being worked on and hopefully should have something within the month.
Sorry, after doing a bit of investigation on this I hadn't gotten back to be active on it. I had just done more investigation since I wasn't familiar with the codebase. I am still happy to help but will step aside. Happy to help with reviews though if it would help.
Heya @johnSchnake! A lot of the work is done and at this point it's connecting the dots.
I'd love to have it tested (heavily) before we cut our next release so any reviews and testing would be welcomed. :)
Kubernetes-dashboard 1.10.1 (https://raw.githubusercontent.com/kubernetes/dashboard/v1.10.1/src/deploy/recommended/kubernetes-dashboard.yaml) not working with kubernetes on-prem cluster version 1.13.1.
'kubectl logs -n kube-system kubernetes-dashboard-57df4db6b-4h5wr' errors out " Metric client health check failed: the server could not find the requested resource (get services heapster). Retrying in 30 seconds"
What's the work around or it's an open issue which needs a fix?
@Angha Kubernetes v1.13 removed Heapster. Short term, you can still install Heapster to get things working.
Long term, there is already a fix in flight (https://github.com/kubernetes/dashboard/pull/3504) that will be in the next Dashboard release.
any guidance on how this is going or how to test what's been done so far (in case you want more folks testing it)
@aleks-mariusz check https://github.com/kubernetes/dashboard/pull/3504
I'm actually trying to test the container but i'm having some issues with the kubeconfig file import
@aleks-mariusz @MrD2 - https://brookbach.com/2018/10/29/Heapster-on-Kubernetes-1.11.3.html
This requires to move back to heapster (which i don't want to)
The only thing not working in the dashboard are the usage visualization.... And for that you have monitoring. I'm just testing the proposed container to maintain compatibility
有新的进展吗?我打算在接下来选型中抛弃 heapster,使用 metrics server。
Guys, do you have any updates on this?
@jeefy Any updates?
@mouhsinelonly You are free to work on that, nobody is stopping you from it I think.
https://github.com/kubernetes/dashboard/pull/3504 Is ready to be reviewed. Please bring any criticisms (constructive or otherwise) and we'll hash this out. :)
@weiqiang333 对不起,标准语言是英语。如果您不会说英语,请使用翻译服务。 (And sorry for my terrible Chinese... I simply used Google Translate...)
As someone who has deployed dashboard w/ heapster, coming back here and trying to figure out if I still need to use heapster, or if I can use metrics-server is _very_ confusing. The wiki still states you have to use heapster?
@chiefy this has been merged, but not yet released.
@labria Do we know the approximate release date for 2.0?
@leezout see https://github.com/kubernetes/dashboard/milestone/1
any idea on new release date for 2.0?
How's this going?
Check the relases :)
Looks like there's a mistake in https://raw.githubusercontent.com/kubernetes/dashboard/v2.0.0-beta1/aio/deploy/recommended.yaml, the deployment refers to kubernetes-dashboard-head while all other resources use kubernetes-dashboard.
I'm also having trouble running this on 1.13. Looks like the dashboard isn't finding the metrics scraper service so isn't starting.
Ya same issue here..I tried deploying this and it worked. It is not on https though
Anyone faced issue mentioned in https://github.com/kubernetes/dashboard/issues/4088 and fixed?
same issue here on 1.14
Does this problem have been fixed on 1.14?
Metrics support is part of v2 beta. Beta is not part of K8s release, we have to make it final then we can update.
Same issue on kubernetes V1.15.2
Dashboard doesn't support metrics on Kubernetes V1.15.3 🙁
Any details about that? Version used? Configuration details? Without additional information, I am just assuming this is a configuration issue on your end.
Any details about that? Version used? Configuration details? Without additional information, I am just assuming this is a configuration issue on your end.
Kubernets V1.15.3 & Dashboard v1.10.1, I installed metrics successful, After refresh the dashboard page, I can't see the CPU graph. I need to configure something?
I have this issue when there are 'completed' pods in my namespace. When I execute kubectl delete pods --field-selector=status.phase=Succeeded --all-namespaces, everything is ok.
@BskyRui what part of the sentence "Metrics support is part of v2 beta" is not clear? It will not work with v1. Also, there are still some issues related to the graphs itself: https://github.com/kubernetes/dashboard/milestone/1
kubeadm v1.16.2 is there any idea how to run dashboard there ?
@patsadow2 We've gotten the v2 beta dashboard running easily in our 1.16.2 cluster, using the manifest provided here.
Most helpful comment
I can start working on this.