Aliaksandr Ivanou

File linter and components finder modules reside under `torchx.specs` module. Components finder module has runtime dependency on the `torchx.components` module: during `find` execution, components finder would dynamically load the module...

Use Meta networking lib (#75894)

4

Summary: X-link: https://github.com/pytorch/pytorch/pull/75894 Move `get_host_addr` to separate module, and implement meta-internal `get_host_addr` that uses meta networking lib. More info: https://docs.google.com/document/d/1paDoXWOy1ws2i1riFypJ3hpP6rO7cPwzTAHnH8OkPKY Reviewed By: kiukchung, Babar Differential Revision: D35288985

Add resource to `utils.sh` component

2

Summary: Add resource to `utils.sh` component Differential Revision: D30563066

dummy change

2

Test plan:

aivanou build time

2

Test plan:

[kubernetes] Pods with wrong image getting stuck forever

6

## 🐛 Bug Module (check all that applies): * [ ] `torchx.spec` * [ ] `torchx.component` * [ ] `torchx.apps` * [ ] `torchx.runtime` * [ ] `torchx.cli` * [x]...

## Description The feature changes the way we perform components integ tests. Instead of manually adding `ComponentsProvider` for each builtin, we find all components available in the code and test...

User loss of work if the cluster change occurs in the middle of the epoch

1

## Description Currently, when the cluster membership change occurs, the agent will kill all the workers that run users script, perform rank redistribution and spawn them again. Since there is...