[
https://issues.apache.org/jira/browse/YUNIKORN-2010?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17777335#comment-17777335
]
Yu-Lin Chen edited comment on YUNIKORN-2010 at 10/19/23 3:53 PM:
-----------------------------------------------------------------
PR updated,
Resuming was cleanup.
Below is all ‘Running Application’ metric changes in application_state. (queue,
queue metric, scheduler metric)
# enter Starting → increment
# leave Completing → increment if destination is Running
# leave Starting → decrement, unless destination is Running
# leave Running → decrement
And I found that ‘Accepted Applications’ was incorrect, so I fixed it in the
same PR:
* {color:#172b4d}Was{color}: leave New → increment
* {color:#172b4d}Is{color}: leave New → increment if destination is not
Reject
was (Author: yu-lin chen):
PR updated,
Resuming was cleanup.
Below is all ‘Running Application’ metric changes in application_state. (queue,
queue metric, scheduler metric)
# enter Starting → increment
# leave Completing → increment if destination is Running
# leave Starting → decrement, unless destination is Running
# leave Running → decrement
And I found that ‘Accepted Applications’ was incorrect, so I fixed it in the
same PR:
* {color:#172b4d}Was{color}: leave New → increment
* {color:#172b4d}Is{color}: leave New → ncrement if destination is not
Reject
> Yunikorn web shows negative number of applications in application history
> -------------------------------------------------------------------------
>
> Key: YUNIKORN-2010
> URL: https://issues.apache.org/jira/browse/YUNIKORN-2010
> Project: Apache YuniKorn
> Issue Type: Bug
> Components: core - scheduler
> Reporter: Yu-Lin Chen
> Assignee: Yu-Lin Chen
> Priority: Major
> Labels: pull-request-available
> Attachments: Nagative Running Application Count - UI Screenshot.png,
> YUNIKORN2010-Application State.png
>
>
> {*}Observed{*}: Please refer to the attached screenshot.
> {*}Root Cause{*}:
> * When application state[1] in core transite from Completing to Running, we
> should increase the total running application count. (Please refer to
> application state flow in attachment.)
> {*}Reproduce Steps{*}:
> 1. Create trigger_sleep_pods.sh
> {code:java}
> #!/bin/bash
> kubectl delete -f sleeppods.yamlwhile true; do
> kubectl apply -f sleeppods.yaml
> sleep 5
> kubectl delete -f sleeppods.yaml
> done{code}
> 2. Create sleeppods.yaml
> {code:java}
> apiVersion: v1
> kind: Pod
> metadata:
> labels:
> app: sleep
> applicationId: "application-sleep-0001"
> queue: "root.sandbox"
> name: task-sleep
> spec:
> schedulerName: yunikorn
> restartPolicy: Never
> containers:
> - name: sleep-300s
> image: "alpine:latest"
> command: ["sleep", "300"]
> resources:
> requests:
> cpu: "100m"
> memory: "500M" {code}
> 3. bash trigger_sleep_pods.sh
> {*}Will fix{*}:
> # Increase running application count when
> ** Completing -> Running
> # We should also decrease running application count when
> ** Starting -> Failing
>
> [1]
> [https://yunikorn.apache.org/docs/next/design/scheduler_object_states/#application-state]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
