Comments (14)
Closing this for now, as the Getting Started
works with a pre rc0
setup:
- Cloning the Gallery and switching to the commit right before the
rc0
compatibility commit:418e93582145c4b07c2c63b11cd00b1061e674fa
(bentoml/gallery@418e935) - In
quickstart/requirements.txt
, specify:bentoml==1.0.0-a7
(https://github.com/bentoml/BentoML/tree/v1.0.0-a7) - Deploy the compatible version of
yatai
using:helm install yatai yatai/yatai -n yatai-system --create-namespace --version v0.3.2
from yatai.
Hi @gaspard-met, in order to setup minikube locally, you need to setup postgres and a minio locally. Make sure to update those values under yatai-config.dev.yaml
. Then try the installation again with the helm chart
from yatai.
@gaspard-met We've discovered a few compatibility issues with 1.0rc with yatai that are going to take some time sort through. If you'd like to work through these issues with us we'd appreciate it, but if you'd like to get something deployed, I'd use 1.0.0a7 just to see it working end to end.
I probably would not downgrade to 0.13. It would require a lot more customization to get it working with yatai I think.
from yatai.
Additional info: upon consulting the UI of yatai
, I am greeted with the same message
ToModelSchemas: GetImageName: cannot get ingress yatai-docker-registry: ingresses.networking.k8s.io "yatai-docker-registry" not found
from yatai.
More on the matter after some research
- I tried the install with a Custom Docker Registry as mentioned in the admin guide
- I ran
minikube delete
- Ran this command
helm install yatai yatai/yatai \
--set config.docker_registry.server='https://index.docker.io/v1' \
--set config.docker_registry.username='MY_DOCKER_USER' \
--set config.docker_registry.password='MY_DOCKER_USER_PASSWORD' \
--set config.docker_registry.secure=true \
-n yatai-system --create-namespace
- The helm install worked
- I tried to push a bento again and first got
Error: [cli] `push` failed: request failed with status code 400: {"error":"pre sign s3 upload url: get bucket yatai exist: Server not initialized, please try again."}
2022-05-27T17:30:12.162149100Z error checking push permissions -- make sure you entered the correct tag name, and that you are authenticated correctly, and try again: checking push permission for "docker.io/yatai-models:yatai.default.iris_clf.srm6ldg54kwcmaav": POST https://index.docker.io/v2/library/yatai-models/blobs/uploads/: UNAUTHORIZED: authentication required; [map[Action:pull Class: Name:library/yatai-models Type:repository] map[Action:push Class: Name:library/yatai-models Type:repository]]
- also the bento push failed :
[2022-05-27 19:30:10] [Pulling] Pulling image "gcr.io/kaniko-project/executor:latest"
[2022-05-27 19:30:10] [Scheduled] Successfully assigned yatai-builders/yatai-bento-image-builder-default-iris-classifier-urg-3-iag-54-khg-2-aav-ca-8-gl-8-l-3-lscja-6-ep-7-teg to minikube
[2022-05-27 19:30:11] [Pulled] Successfully pulled image "gcr.io/kaniko-project/executor:latest" in 444.8169ms
[2022-05-27 19:30:11] [Started] Started container builder
[2022-05-27 19:30:11] [Created] Created container builder
[2022-05-27 19:30:15] [FailedMount] MountVolume.SetUp failed for volume "kube-api-access-9zmv8" : object "yatai-builders"/"kube-root-ca.crt" not registered
[2022-05-27 19:30:15] [FailedMount] MountVolume.SetUp failed for volume "docker-config" : object "yatai-builders"/"docker-config" not registered
from yatai.
Hi @aarnphm ,
Thanks for the answer.
Do you mean to update the yaml in the Yatai repo here https://github.com/bentoml/Yatai/blob/main/yatai-config.sample.yaml
or in the Yatai Chart repo here https://github.com/bentoml/yatai-chart/blob/main/values.yaml ?
Also, no mention is made of minio in either of them, should it be added ?
Do you have an example of what the yaml file looks like with postgres
and minio
? postgres seems to be active in the helm chart for yatai by default (https://github.com/bentoml/yatai-chart/blob/5a7272bdc6294303e226f09ad12efc45f793d27f/values.yaml#L34)
I am mistaken or is this extra setup necessary to execute the Getting Started example from the readme ?
If it is, them wouldn't it be easier to have these values by default or more documentation in the readme example directly ?
from yatai.
are you installing yatai from the helm chart or from the git repo?
from yatai.
Via the Helm Chart (exactly as listed in the Getting Started example)
I looked at values.yaml
, and postgresql
is enabled
minio
isn't mentioned there directly, maybe it corresponds to a specific service ?
from yatai.
Looking at the services running on Minikube, minio
seems to be there:
from yatai.
Hey @aarnphm, still cannot make the Yatai example work on my side.
Using the latest of everything locally with Minikube
I manage now to push the models to Yatai
- Logs from model upload (success)
2022-05-31T12:34:57.522774500Z INFO[0000] No base image, nothing to extract
2022-05-31T12:34:57.522964500Z INFO[0000] Built cross stage deps: map[]
2022-05-31T12:34:57.522975400Z INFO[0000] No base image, nothing to extract
2022-05-31T12:34:57.522979300Z INFO[0000] Executing 0 build triggers
2022-05-31T12:34:57.523441000Z INFO[0000] Unpacking rootfs as cmd COPY . /model requires it.
2022-05-31T12:34:57.523522500Z INFO[0000] COPY . /model
2022-05-31T12:34:57.524046800Z INFO[0000] Taking snapshot of files...
2022-05-31T12:34:57.528120900Z INFO[0000] Pushing image to yatai-docker-registry-yatai-infra-cluster-10-105-66-207.apps.yatai.dev/models:yatai.default.iris_clf.au24kaxa32d7eaav
2022-05-31T12:34:57.592041800Z INFO[0000] Pushed yatai-docker-registry-yatai-infra-cluster-10-105-66-207.apps.yatai.dev/models@sha256:[sha removed here]
- Events from model upload
[2022-05-31 14:34:55] [Scheduled] Successfully assigned yatai-builders/yatai-model-image-builder-default-iris-clf-au-24-kaxa-32-d-7-eaav-cab-0-mrof-573-i-1-fhbrq-9-g to minikube
[2022-05-31 14:34:56] [Pulling] Pulling image "gcr.io/kaniko-project/executor:latest"
[2022-05-31 14:34:57] [Pulled] Successfully pulled image "gcr.io/kaniko-project/executor:latest" in 536.2571ms
[2022-05-31 14:34:57] [Started] Started container builder
[2022-05-31 14:34:57] [Created] Created container builder
[2022-05-31 14:35:00] [FailedMount] MountVolume.SetUp failed for volume "docker-file-3" : object "yatai-builders"/"docker-file-3" not registered
[2022-05-31 14:35:00] [FailedMount] MountVolume.SetUp failed for volume "docker-config" : object "yatai-builders"/"docker-config" not registered
[2022-05-31 14:35:00] [FailedMount] MountVolume.SetUp failed for volume "kube-api-access-9hcv9" : object "yatai-builders"/"kube-root-ca.crt" not registered
And then however the bento build fails :
- Logs from bento build (fails)
2022-05-31T12:34:57.732637600Z error building image: parsing dockerfile: dockerfile parse error line 85: unknown instruction: IF
- Events from bento build
[2022-05-31 14:34:56] [Scheduled] Successfully assigned yatai-builders/yatai-bento-image-builder-default-iris-classifier-bq-42-lmxa-32-qsuaav-cab-0-ms-0-f-573-i-1-fhbrqa-0 to minikube
[2022-05-31 14:34:56] [Pulling] Pulling image "gcr.io/kaniko-project/executor:latest"
[2022-05-31 14:34:57] [Pulled] Successfully pulled image "gcr.io/kaniko-project/executor:latest" in 769.6544ms
[2022-05-31 14:34:57] [Started] Started container builder
[2022-05-31 14:34:57] [Created] Created container builder
[2022-05-31 14:35:00] [FailedMount] MountVolume.SetUp failed for volume "docker-config" : object "yatai-builders"/"docker-config" not registered
[2022-05-31 14:35:00] [FailedMount] MountVolume.SetUp failed for volume "kube-api-access-cw4jv" : object "yatai-builders"/"kube-root-ca.crt" not registered
I have started from a clean Minikube several times, and to the best of my knowledge it's exactly the config as supposed to be used in the Getting Started example of Yatai.
Have you managed to reproduce this ?
thanks
from yatai.
Here is a sample, with the culprit at line 85 highlighted, from the latest
bento Dockerfile found in ~/bentoml/bentos/iris_classifiers
:
from yatai.
This is due to our recent refactor on docker build process to support multiple architecture. A update on Yatai should come out soon.
from yatai.
Should I downgrade to an alpha release or even to 0.13 LTS ?
from yatai.
Hi @timliubentoml ,
Thanks for the tip, I will try it out 👍
If it works with 1.0.0a7
I'll close the issue for now
P.S.: I can give you a hand to the extent of my limited knowledge on the topic to get the issues fixed 😄
from yatai.
Related Issues (20)
- fix trans in BentoCloud HOT 1
- Error: kaniko(bentoImageBuildEngine) apt-get install error HOT 1
- Error: get Bento CR bentoes.resources.yatai.ai not found HOT 1
- Will yatai login ldap be supported? HOT 1
- Deploy a "custom Bentoml image" using "yatai-deployment".
- Support labels from values.yaml
- Error: YAML parse error on yatai/templates/deployment.yaml HOT 4
- Installation of yatai-deployment failed to [ERROR registerYataiComponent...] HOT 5
- Provide a way to disable home page cards
- Cannot access web ui HOT 1
- can not start dashboard when running yarn start HOT 3
- Autoscaling LLM HOT 2
- how to deploy with cutom-runner HOT 1
- Does yatai deployment require saving the model to the bentos model store? HOT 2
- Yatai 2.0 Proposal HOT 1
- Support BentoML 1.2 in Yatai HOT 1
- How can I deploy bentoDeployment with both remote runner and embedded runner?
- How can I host the yatai image in a private registry HOT 3
- Yatai memory leak HOT 7
- Request to set the root path of the Dashboard
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from yatai.