[
https://issues.apache.org/jira/browse/YUNIKORN-406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17195197#comment-17195197
]
Akhil PB edited comment on YUNIKORN-406 at 9/14/20, 6:29 AM:
-------------------------------------------------------------
Hi [~vbm] [~wwei],
Ingress seems good idea too. If we do configure ingress for only the UI along
with nginx proxy, I think we can get away with port forwarding of both ports.
I have couple of ideas.
*Idea 1:* Configure ingress for UI (ingress path could be like
http://<INGRESS_IP>/yunikorn) and by knowing ingress IP, user can hit
{{http://<INGRESS_IP>/yunikorn}}. The ingress will redirect the traffic to
nginx server (server hosting web UI). All the web assets will be delivered by
the server. With the api proxy configured in nginx to talk to YK scheduler, UI
could call api too without any port forwarding. [No need to edit envconfig.json
file]
*Idea 2:* Only with nginx proxy and port forwarding only the web UI port, we
could achieve the same. Here we are not configuring any ingress. [No need to
edit envconfig.json file]
Please share your thoughts. Also note that we may have to spent some R&D time
on these to verify how feasible these ideas are. I strongly feel idea-2 is
feasible, but I don't know much about idea-1, have to spend some time with
ingress in a YK deployed cluser.
was (Author: akhilpb):
Hi [~vbm] [~wwei],
Ingress seems good idea too. If we do configure ingress for only the UI along
with nginx proxy, I think we can get away with port forwarding of both ports.
I have couple of ideas.
*Idea 1:* Configure ingress for UI (ingress path could be like
http://<INGRESS_IP>/yunikorn) and by knowing ingress IP, user can hit
{{http://<INGRESS_IP>/yunikorn}}. The ingress will redirect the traffic to
nginx server (server hosting web UI). All the web assets will be delivered by
the server. With the api proxy configured in nginx to talk to YK scheduler, UI
could call api too without any port forwarding. [No need to edit envconfig.json
file]
*Idea 2:* Only with nginx proxy and port forwarding only the web UI port, we
could achieve the same. Here we are not configuring any ingress. [No need to
edit envconfig.json file]
Please share your thoughts.
> Queue/apps is not getting displayed on yunikorn UI
> --------------------------------------------------
>
> Key: YUNIKORN-406
> URL: https://issues.apache.org/jira/browse/YUNIKORN-406
> Project: Apache YuniKorn
> Issue Type: Bug
> Components: core - scheduler, webapp
> Reporter: Vishwas
> Assignee: Akhil PB
> Priority: Major
> Attachments: access_via_ingress.PNG, error_ui_console.PNG,
> error_ui_default_envconfig.PNG, image-2020-09-14-09-52-50-388.png,
> node_info.PNG, queue_info.PNG
>
>
> Hi,
> I am a newbie in to yunikorn. I am using the recently release 0.9.0 helm
> chart.
> I have disabled embedAdmissionController and below is the queue configuration
> in values.yaml
>
> {code:java}
> # Use this configuration to automatically map K8s namespaces to yunikorn
> queues
> configuration: |
> partitions:
> - name: default
> placementrules:
> - name: tag
> value: namespace
> create: true
> queues:
> - name: root
> submitacl: '*'
> queues:
> - name: tpcds
> {code}
>
>
> i am trying to launch a simple sleep pod just to see if the pod gets
> allocated to proper queue.
>
> {code:java}
> apiVersion: v1
> kind: Pod
> metadata:
> labels:
> app: sleep
> applicationId: "appli-sleepp-0001"
> #queue: "root"
> namespace: tpcds
> name: app-sleep-0
> spec:
> schedulerName: yunikorn
> containers:
> - name: sleep-30s
> image: "alpine:latest"
> command: ["sleep", "inf"]
> resources:
> requests:
> cpu: "100m"
> memory: "500M"
> ---
> {code}
>
>
> I see that the job i getting allocated properly to root.tpcds queue.I am able
> to confirm this only using rest api but I am not able to see anything in
> yunikorn-ui.
>
> With rest-api:
> {code:java}
> curl http://10.75.38.79:31165/ws/v1/apps
> [
> {
> "applicationID": "appli-sleepp-0001",
> "usedResource": "[memory:500 vcore:100]",
> "partition": "[mycluster]default",
> "queueName": "root.tpcds",
> "submissionTime": 1599644324773451800,
> "allocations": [
> {
> "allocationKey": "f095ab0f-cc25-40d2-8889-2dba31f51441",
> "allocationTags": null,
> "uuid": "83d6acbe-3242-4321-9207-fd6c6d4f38b5",
> "resource": "[memory:500 vcore:100]",
> "priority": "<nil>",
> "queueName": "root.tpcds",
> "nodeId": "vm-10-75-38-73",
> "applicationId": "appli-sleepp-0001",
> "partition": "default"
> }
> ],
> "applicationState": "Running"
> }
> ]
> {code}
>
> I do not see the same info in yunikorn-ui. I have attached the node-info and
> queue-info from UI.
> I have tried on firefox and chrome and the same issue is seen.
>
> kubernetes version: 1.17
> {code:java}
> [root@vm-10-75-38-79 vbm]# kubectl version
> Client Version: version.Info{Major:"1", Minor:"17", GitVersion:"v1.17.4",
> GitCommit:"8d8aa39598534325ad77120c120a22b3a990b5ea", GitTreeState:"clean",
> BuildDate:"2020-03-12T21:03:42Z", GoVersion:"go1.13.8", Compiler:"gc",
> Platform:"linux/amd64"}
> Server Version: version.Info{Major:"1", Minor:"17", GitVersion:"v1.17.4",
> GitCommit:"8d8aa39598534325ad77120c120a22b3a990b5ea", GitTreeState:"clean",
> BuildDate:"2020-03-12T20:55:23Z", GoVersion:"go1.13.8", Compiler:"gc",
> Platform:"linux/amd64"}
> {code}
> {code:java}
> [root@vm-10-75-38-79 vbm]# kubectl get all -n yunikorn
> NAME READY STATUS RESTARTS AGE
> pod/yunikorn-scheduler-59c7657ccb-lktxp 2/2 Running 0
> 4h1mNAME TYPE CLUSTER-IP EXTERNAL-IP
> PORT(S) AGE
> service/yunikorn-service NodePort 10.254.242.166 <none>
> 9080:31165/TCP,9889:30840/TCP 4h1mNAME
> READY UP-TO-DATE AVAILABLE AGE
> deployment.apps/yunikorn-scheduler 1/1 1 1
> 4h1mNAME DESIRED CURRENT READY
> AGE
> replicaset.apps/yunikorn-scheduler-59c7657ccb 1 1 1
> 4h1m
> {code}
>
>
>
>
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
