Page MenuHomePhabricator
Paste P96467

Issues on deploying the Qwen3.8-27b model on llm namespace
ActivePublic

Authored by gkyziridis on Thu, Sep 17, 11:25 AM.
Project Tags
None
Referenced Files
F103431988: Issues on deploying the Qwen3.8-27b model on llm namespace
Thu, Sep 17, 11:25 AM
Subscribers
None
$ kubectl get isvc llm-qwen38-27b
NAME URL READY PREV LATEST PREVROLLEDOUTREVISION LATESTREADYREVISION AGE
llm-qwen38-27b False 27m
$ kubectl get ksvc llm-qwen38-27b-predictor
NAME URL LATESTCREATED LATESTREADY READY REASON
llm-qwen38-27b-predictor http://llm-qwen38-27b-predictor.llm.wikimedia.org llm-qwen38-27b-predictor-00001 llm-qwen38-27b-predictor-00001 True
$ kubectl get revision | grep qwen38
llm-qwen38-27b-predictor-00001 llm-qwen38-27b-predictor 1 True 1 1
$ kubectl logs llm-qwen38-27b-predictor-00001-deployment-5d77fdc4b5-lw5vn -c kserve-container --tail=5
INFO:kserve.trace:kserve.io.metrics: 0.004782199859619141 ['http_status:200', 'http_method:GET', 'time:wall']
INFO:kserve.trace:kserve.io.metrics: 0.004768999999999579 ['http_status:200', 'http_method:GET', 'time:cpu']
INFO:uvicorn.access:127.0.0.6:53513 - "GET /metrics HTTP/1.1" 200
INFO:kserve.trace:kserve.io.metrics: 0.004786491394042969 ['http_status:200', 'http_method:GET', 'time:wall']
INFO:kserve.trace:kserve.io.metrics: 0.0047739999999976135 ['http_status:200', 'http_method:GET', 'time:cpu']
$ kubectl describe isvc llm-qwen38-27b | tail -40
Severity: Info
Status: Unknown
Type: PredictorRouteReady
Last Transition Time: 2026-09-17T10:50:55Z
Message: fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationkbx84" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Reason: ReconcileFailed
Status: False
Type: Ready
Last Transition Time: 2026-09-17T10:29:00Z
Reason: PredictorRouteReady not ready
Severity: Info
Status: Unknown
Type: RoutesReady
Last Transition Time: 2026-09-17T10:29:00Z
Severity: Info
Status: False
Type: Stopped
Deployment Mode: Knative
Model Status:
Copies:
Failed Copies: 0
States:
Active Model State:
Target Model State: Pending
Transition Status: InProgress
Observed Generation: 1
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationzbsds" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validation8scrg" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationm5vpt" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validation24qv2" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationtlsrj" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationlgjsz" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationdwzmm" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationqrp2f" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning InternalError 21m v1beta1Controllers fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validation44fqh" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template
Warning UpdateFailed 21m (x2 over 21m) v1beta1Controllers Failed to update status for InferenceService "llm-qwen38-27b": Operation cannot be fulfilled on inferenceservices.serving.kserve.io "llm-qwen38-27b": the object has been modified; please apply your changes to the latest version and try again
Warning InternalError 115s (x5947 over 21m) v1beta1Controllers (combined from similar events): fails to reconcile predictor: fails to reconcile knative service: admission webhook "validation.webhook.serving.knative.dev" denied the request: validation callback failed: dry run failed with pods "dry-run-validationgx479" is forbidden: exceeded quota: quota-compute-resources, requested: limits.cpu=18, used: limits.cpu=145, limited: limits.cpu=150: spec.template