[
https://issues.apache.org/jira/browse/YUNIKORN-3121?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wilfred Spiegelenburg updated YUNIKORN-3121:
--------------------------------------------
Target Version: 1.9.0 (was: 1.8.0)
> Add REST API Endpoint for Scheduling Order Visibility
> -----------------------------------------------------
>
> Key: YUNIKORN-3121
> URL: https://issues.apache.org/jira/browse/YUNIKORN-3121
> Project: Apache YuniKorn
> Issue Type: New Feature
> Components: core - scheduler
> Reporter: Mit Desai
> Assignee: Mit Desai
> Priority: Major
> Labels: pull-request-available
>
> h3. Summary
> Add a new REST API endpoint that provides real-time visibility into the
> evaluation order of queues and applications during YuniKorn scheduling
> cycles. This enhancement will significantly improve debugging capabilities by
> allowing operators to quickly identify which queues and applications are
> being processed by the scheduler and in what order.
> h3. Background
> When debugging YuniKorn scheduling performance issues, operators often need
> to understand which queues and applications are being evaluated during
> scheduling cycles. Currently, this information is not easily accessible,
> requiring operators to:
> * Guess which applications might be causing scheduling delays
> * Search through long lists of applications and queues without clear
> prioritization
> * Analyze complex log files to understand scheduling behavior
> * Spend significant time identifying the root cause of scheduling latency
> This lack of visibility makes it challenging to quickly pinpoint applications
> or queues that might be monopolizing scheduler resources or causing
> performance bottlenecks.
> Add a new REST API endpoint that exposes the current scheduling evaluation
> order, showing:
> # {*}Queue Evaluation Order{*}: The sequence in which queues are being
> processed
> # {*}Application Order per Queue{*}: The order of applications within each
> queue
> # {*}Real-time Data{*}: Current state reflecting the scheduler's evaluation
> logic
> # {*}Structured Format{*}: Easy-to-parse JSON response for both human
> operators and automated tools
> h4. API Specification:
> {*}Endpoint{*}: {{GET /ws/v1/partition/\{partitionName}/schedulingorder}}
> {*}Response Format{*}:
> {noformat}
> [
> {
> "queueName": "root.production.high-priority",
> "applicationIDs": [
> "app-critical-workload-001",
> "app-critical-workload-002"
> ]
> },
> {
> "queueName": "root.production.default",
> "applicationIDs": [
> "app-batch-job-123",
> "app-data-processing-456"
> ]
> },
> {
> "queueName": "root.development",
> "applicationIDs": [
> "app-test-environment-789"
> ]
> }
> ]
> {noformat}
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
