Simon_CQK

```python class Storage(object): SUPERBLOCK_SIZE = 4096 INTEGER_FORMAT = "!Q" INTEGER_LENGTH = 8 ``` I know the INTEGER_FORMAT , but the other two really confuse me.

Set user customized wait-timeout-seconds for PodGroup based gang scheduling protocol.

6

#### What would you like to be added: Introduce a new filed in `podGroup.spec` named with `waitTimeoutSeconds` or sth similar, then users are able to configure `waitTimeoutSeconds` dynamically. #### Why...

# KubeDL 2022 Annual Review ## Table of Contents ## Backgroud KubeDL is a suite of Kubernentes controllers that enable running machine learning workloads on Kubernentes, such as model training...

refactor(mpi): remove global launcherRunsWorkload flag.

9

Signed-off-by: SimonCqk ### Ⅰ. Describe what this PR does remove `launcherRunsWorkload` global startup flag, which can be inferred by whether has `Launcher` role in job spec. ### II. Does this...

[feature request] bump controller-runtime vendor version to k8s 1.20

1

**What would you like to be added**: In k8s `>=1.20`, align with `controller-runtime >=0.8.x`, list/watch transport protocol of built-in resources(pods, services...) has upgraded from `json` to `protobuf`, which brings considerable...

[feature request] job level scheduling and orchestration

2

**What would you like to be added**: 1. a job-level queuing and orchestration module that admit job to be scheduled by some strategies. 2. refactor skeleton codes to make control...

🧑‍💻 🏕 Alibaba Summer of Code (ASOC) 2022

3

# Alibaba Summer of Code (ASOC) 2022 Welcome to the open source world! If you haven't planned how to spend this summer, come to the Alibaba Summer of Code and...

**Background**: As the [official portal](https://pytorch.org/docs/stable/distributed.elastic.html) introduced, torch-elastic has been upstreamed to pytorch >=1.9. KubeDL manages the lifecycle of jobs and orchestrate their resources, it is critical to implement torch-elastic distributed...

**What would you like to be added**: the logging in stdout is mixed with multiple formats for we using logging libraries mixture across different packages and controller implementations, consolidating loggers...

**What would you like to be added**: as we discussed in https://github.com/kubedl-io/kubedl/issues/180#issuecomment-942907449, the global flag launcherRunsWorkload can be removed, mpiReplicaSpecs with Launcher role indicates that mpijob will be driven by...