Wei Huang created YUNIKORN-2031:
-----------------------------------
Summary: metric yunikorn_root_.*_resource{state="pending"} is
inaccurate
Key: YUNIKORN-2031
URL: https://issues.apache.org/jira/browse/YUNIKORN-2031
Project: Apache YuniKorn
Issue Type: Bug
Affects Versions: 1.3.0, 1.2.0
Reporter: Wei Huang
Given a sample static queue configmap yaml:
{code:java}
apiVersion: v1
data:
queues.yaml: |
partitions:
- name: default
queues:
- name: root
submitacl: '*'
queues:
- name: dev
resources:
max:
memory: 20Gi
vcore: "5"
- name: sre
resources:
max:
memory: 20Gi
vcore: "5"
kind: ConfigMap
metadata:
name: yunikorn-configs
namespace: default {code}
and a Pod template yaml like:
{code:java}
kind: Pod
apiVersion: v1
metadata:
generateName: dev-
labels:
applicationId: dev
queue: root.dev
spec:
schedulerName: yunikorn
containers:
- name: pause
image: registry.k8s.io/pause
resources:
requests:
cpu: 1
limits:
cpu: 1 {code}
If I create 6 pods:
{code:java}
for i in {1..6}; do k create -f pods/dev-pod.yaml; done {code}
5 Pods are in Running state, 1 is in Pending state. Which looks good as dev
queue's max quota is 5 cpus.
However, if I check the metric yunikorn_root_dev_queue_resource, it shows
incorrect pending pods/resource:
||Metric||Value||
|yunikorn_root_dev_queue_resource{instance="localhost:9080", job="yunikorn",
resource="pods", state="allocated"}|5|
|yunikorn_root_dev_queue_resource{instance="localhost:9080", job="yunikorn",
resource="pods", state="pending"}|2|
|yunikorn_root_dev_queue_resource{instance="localhost:9080", job="yunikorn",
resource="vcore", state="allocated"}|5000|
|yunikorn_root_dev_queue_resource{instance="localhost:9080", job="yunikorn",
resource="vcore", state="pending"}|2000|
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
