Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix wait-for-unzip race #18

Merged
merged 4 commits into from
Sep 21, 2022
Merged

Fix wait-for-unzip race #18

merged 4 commits into from
Sep 21, 2022

Conversation

knighton
Copy link
Contributor

Fix two identical race conditions in shard preload and download respectively.

The problem: some workers were thinking they were all clear to go ahead before that was actually the case, and encountering momentarily missing, empty, or truncated shards as a result in scenarios involving enough GPUs, large enough shards, and compression. Streaming V2 handles compression differently than V1 and adds hash (and ideally file size) validation, but the "wait" path was not updated to include the added steps in prepping shards for use.

streaming/base/dataset.py Show resolved Hide resolved
streaming/base/dataset.py Show resolved Hide resolved
streaming/base/dataset.py Show resolved Hide resolved
Copy link
Collaborator

@karan6181 karan6181 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Thanks!

@karan6181 karan6181 merged commit df01240 into main Sep 21, 2022
@karan6181 karan6181 deleted the james/wait-for-unzip branch September 21, 2022 22:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants