[BUG] Pull pre Image failed on K8S cluster

[JackyYangPassion]

Describe the bug
Pull pre Image failed on K8S cluster

To Reproduce
Steps to reproduce the behavior:

1. install graphscope pre

pip3 install -U graphscope-client --pre   -i http://mirrors.aliyun.com/pypi/simple/ --trusted-host mirrors.aliyun.com

launch k8s

import graphscope
graphscope.set_option(log_level='DEBUG')
graphscope.set_option(show_log=True)
# Create GraphScope client session, the 'cluster_type' is k8s by default.
session = graphscope.session(
                             k8s_coordinator_cpu=1,
                             k8s_coordinator_mem="1Gi",
                             k8s_vineyard_cpu=0.2,
                             k8s_vineyard_mem="1Gi",
                             vineyard_shared_mem="1Gi",
                             k8s_engine_cpu=0.2,
                             k8s_namespace='gs-new-orc-jacky',
                             k8s_engine_mem="1Gi",
                             num_workers=1,
                             enabled_engines="analytical,interactive",
                             k8s_client_config='~/.kube/config')
print('========= Session created. ==========')

See error

2023-11-20 16:12:26,603 [INFO][utils:187]: coordinator-jzwbkf-65fb74466f-nb967: Failed to pull image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119": Error response from daemon: manifest for registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119 not found: manifest unknown: manifest unknown
2023-11-20 16:12:26,604 [INFO][utils:187]: coordinator-jzwbkf-65fb74466f-nb967: Error: ErrImagePull
2023-11-20 16:12:26,605 [INFO][utils:187]: coordinator-jzwbkf-65fb74466f-nb967: Back-off pulling image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119"
coordinator-jzwbkf-65fb74466f-nb967: Successfully assigned gs-new-orc-jacky/coordinator-jzwbkf-65fb74466f-nb967 to minikube
coordinator-jzwbkf-65fb74466f-nb967: Pulling image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119"
coordinator-jzwbkf-65fb74466f-nb967: Failed to pull image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119": Error response from daemon: manifest for registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119 not found: manifest unknown: manifest unknown
coordinator-jzwbkf-65fb74466f-nb967: Error: ErrImagePull
coordinator-jzwbkf-65fb74466f-nb967: Back-off pulling image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231119"

Environment (please complete the following information):

• GraphScope version: 0.26.0a20231119
• OS: MacOS
• Kubernetes Version 1.23

[siyuan0322]

Recent 3 image building failed. The latest usable image are 0.26.0a20231116, please specify it manually by k8s_image_tag='0.26.0a20231116'.
We will fix the building process ASAP.

For reference: https://github.com/alibaba/GraphScope/actions/workflows/build-graphscope-images-linux.yml

[JackyYangPassion]

Recent 3 image building failed. The latest usable image are 0.26.0a20231116, please specify it manually by k8s_image_tag='0.26.0a20231116'. We will fix the building process ASAP.

For reference: alibaba/GraphScope/actions/workflows/build-graphscope-images-linux.yml

Thanks for your reply.

Motivation:
For this bug fix #3363, I want to use this PR on the PRD env ASAP.

I think of two ways

1. Pull the official pre version, because it is built and released every day
2. Compile locally and publish to internal docker image repo

Can you guide me how to publish the image from the local development environment to the internal repo?

[siyuan0322]

I'll first try to rerun it, seems it stuck during trying to apt update.

You could reference to this workflow file and I think you only need to

1. Rerun make analytical in your case.
2. Tag it appropriate tag registry.cn-hongkong.aliyuncs.com/graphscope/analytical:tag
3. If your cluster is a single machine cluster like minikube, you use minikube image load xxx to load it into your control-plane.
4. If your cluster is not a single machine, you could docker save and docker load the image to the nodes machine.
5. And make sure the image pull policy is IfNotPresent.

[siyuan0322]

Fixed by #3365 , The images would be available by tomorrow.

[JackyYangPassion]

coordinator-mvekpb-5f9ffddcdd-4nck8: Failed to pull image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231121": rpc error: code = Unknown desc = Error response from daemon: manifest for registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231121 not found: manifest unknown: manifest unknown
2023-11-22 16:03:20,061 [INFO][utils:187]: coordinator-mvekpb-5f9ffddcdd-4nck8: Error: ErrImagePull
coordinator-mvekpb-5f9ffddcdd-4nck8: Successfully assigned gs-new-orc-jacky2/coordinator-mvekpb-5f9ffddcdd-4nck8 to
coordinator-mvekpb-5f9ffddcdd-4nck8: Pulling image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231121"
coordinator-mvekpb-5f9ffddcdd-4nck8: Failed to pull image "registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231121": rpc error: code = Unknown desc = Error response from daemon: manifest for registry.cn-hongkong.aliyuncs.com/graphscope/coordinator:0.26.0a20231121 not found: manifest unknown: manifest unknown

[JackyYangPassion]

Pull pre Image failed on K8S cluster still error @siyuan0322

[siyuan0322]

Sorry, the disk space of runner is exhausted. Rerunning.

[JackyYangPassion]

Thanks for your reply, I will try it tomorrow

[siyuan0322]

This tag is available now 0.26.0a20231122