[PR #1057] [MERGED] handling cudaMemcpy errors in tcp_transport.cpp #1984

New issue

Closed

opened 2026-10-29 21:31:53 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-29 21:31:53 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/Mooncake/pull/1057
Author: @flying-x
Created: 11/13/2025
Status: ✅ Merged
Merged: 11/14/2025
Merged by: @stmatengss

Base: main ← Head: main

📝 Commits (4)

9d3b02d handling cudaMemcpy errors in tcp_transport.cpp
f09e4b7 Update mooncake-transfer-engine/src/transport/tcp_transport/tcp_transport.cpp
cc9d81e Update mooncake-transfer-engine/src/transport/tcp_transport/tcp_transport.cpp
4b233f2 format

📊 Changes

1 file changed (+25 additions, -4 deletions)

View changed files

📝 mooncake-transfer-engine/src/transport/tcp_transport/tcp_transport.cpp (+25 -4)

📄 Description

when cuda memory is almost full, cudaMemcpy can fail
without the error checking, the transfer engine swallow the error, which resulted in upper level applications to behave incorrectly.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/kvcache-ai/Mooncake/pull/1057 **Author:** [@flying-x](https://github.com/flying-x) **Created:** 11/13/2025 **Status:** ✅ Merged **Merged:** 11/14/2025 **Merged by:** [@stmatengss](https://github.com/stmatengss) **Base:** `main` ← **Head:** `main` --- ### 📝 Commits (4) - [`9d3b02d`](https://github.com/kvcache-ai/Mooncake/commit/9d3b02d709e0c7cabe6a1fdf59e8d5bfeb090efa) handling cudaMemcpy errors in tcp_transport.cpp - [`f09e4b7`](https://github.com/kvcache-ai/Mooncake/commit/f09e4b7c521563336def0a99e18107f9d9cd0a11) Update mooncake-transfer-engine/src/transport/tcp_transport/tcp_transport.cpp - [`cc9d81e`](https://github.com/kvcache-ai/Mooncake/commit/cc9d81e43b160542c8c45a140fc8b15f8e413f4e) Update mooncake-transfer-engine/src/transport/tcp_transport/tcp_transport.cpp - [`4b233f2`](https://github.com/kvcache-ai/Mooncake/commit/4b233f2a4382576daad5e2db1885fd5de926ac16) format ### 📊 Changes **1 file changed** (+25 additions, -4 deletions) <details> <summary>View changed files</summary> 📝 `mooncake-transfer-engine/src/transport/tcp_transport/tcp_transport.cpp` (+25 -4) </details> ### 📄 Description - when cuda memory is almost full, cudaMemcpy can fail - without the error checking, the transfer engine swallow the error, which resulted in upper level applications to behave incorrectly. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-29 21:31:53 +00:00