12 May 2022 |
Mark Winter | https://velog.io/@pjs102793/Triton-Inference-Server%EC%97%90%EC%84%9C-TensorRT-Engine-Inference | 06:22:44 |
Mark Winter | Triton 서버를 동작시키기 위해서는 반드시 모델 파일과 모델 파일에 대한 Configuration 들을 포함하고 있는 Model Repository 를 생성해야 합니다. | 06:25:05 |
zorba(손주형) | oh i see. it’s just same as usning triton. Thanks! | 06:25:36 |
Mark Winter | Yer, just set modelFormat to tensorrt and you will get a Triton server | 06:26:49 |
Mark Winter | Would there be any issues with having serverless kserve and modelmesh in the same cluster? | 06:31:32 |
13 May 2022 |
| _slack_kubeflow_U03FCED8QTE joined the room. | 05:32:17 |
Saurabh Agarwal | Shri Javadekar ^ | 12:20:47 |
Alexandre Brown | Redacted or Malformed Event | 17:36:35 |
Shri Javadekar | • What version of Kserve are you using? I'm using kserve 0.7 and there is a :predict suffix to the endpoint url (http://my-model-a.kserve-test.svc.cluster.local/v1/models/my-model-a:predict )
• When you say that the url isn't working, what's the error you are seeing? | 20:56:58 |
14 May 2022 |
Dan Sun | Should work fine, that’s all the work we have doing to unifying on the inference service api to deploy single and multi models | 20:06:34 |
15 May 2022 |
| Amanjeet Sahu joined the room. | 08:38:19 |
16 May 2022 |
wenyang zhou | Download 截屏2022-05-16 上午11.04.47.png | 03:06:39 |
wenyang zhou | There seems to be a lock | 03:06:39 |
wenyang zhou | I have release several revision in a period of time. | 03:09:02 |
wenyang zhou | The old revision can not terminate before it finish pending . | 03:09:47 |
wenyang zhou | And the new revision can not allocate enough resource before the old revision release resource. | 03:10:33 |
wenyang zhou | Could we improve this problem ? | 03:10:54 |
wenyang zhou | Current revision is 15 with 100% traffic percent, and I have roll out the revision 16 with 100% this time. | 03:14:19 |
wenyang zhou | After a few minutes, the revision 16 has release resources. But 13 and 17 keep it. | 03:16:32 |
wenyang zhou | Is this a knative problem? | 03:23:19 |
wenyang zhou | Download 截屏2022-05-16 上午11.32.16.png | 03:32:32 |
wenyang zhou | So confused! | 03:32:58 |
wenyang zhou | Revision 13 can not terminate~ | 03:33:31 |
Saurabh Agarwal | Shri Javadekar host not found error | 06:27:47 |
Saurabh Agarwal | v0.7.0 this is the version | 07:36:59 |
| idahotokens joined the room. | 11:11:19 |
| idahotokens left the room. | 16:30:27 |
Cesar Flores | Hi everyone, is anyone using a way to define the output of the models for downstream applications? like for example having openAPI so the consumers of the models know what the ouutput of the response will be??? | 16:30:46 |
Cesar Flores | this is pretty annoying, but if it helps in any way, I solved my issue by using the Host: as the url that appears when you run kubectl get ksvc -n namespace | 16:31:53 |
Alexandre Brown | Saurabh Agarwal Make sure to add the host to the headers | 16:32:43 |