[PR #285] [MERGED] [TransferEngine] only dump metadata on the last selectDevice failure #1451

New issue

Closed

opened 2026-10-29 21:27:33 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-29 21:27:33 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/Mooncake/pull/285
Author: @eryugey
Created: 4/22/2025
Status: ✅ Merged
Merged: 4/23/2025
Merged by: @alogfans

Base: main ← Head: dump

📝 Commits (1)

a46f0e9 only dump metadata on the last selectDevice failure

📊 Changes

1 file changed (+5 additions, -5 deletions)

View changed files

📝 mooncake-transfer-engine/src/transport/rdma_transport/worker_pool.cpp (+5 -5)

📄 Description

We selectDevice() according to peer segment desc, and will force update segmeng desc if this fails, then try selectDevice() for the second time, which normally succeeds.

But currently we dump metadata content on first selectDevice() failure, which is mostly like a transient failure. So let's dump metadata only on the last selectDevice() failure.

_{🔄 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/285 **Author:** [@eryugey](https://github.com/eryugey) **Created:** 4/22/2025 **Status:** ✅ Merged **Merged:** 4/23/2025 **Merged by:** [@alogfans](https://github.com/alogfans) **Base:** `main` ← **Head:** `dump` --- ### 📝 Commits (1) - [`a46f0e9`](https://github.com/kvcache-ai/Mooncake/commit/a46f0e9cc1fffb12d1cac721f18076a5a2d31840) only dump metadata on the last selectDevice failure ### 📊 Changes **1 file changed** (+5 additions, -5 deletions) <details> <summary>View changed files</summary> 📝 `mooncake-transfer-engine/src/transport/rdma_transport/worker_pool.cpp` (+5 -5) </details> ### 📄 Description We selectDevice() according to peer segment desc, and will force update segmeng desc if this fails, then try selectDevice() for the second time, which normally succeeds. But currently we dump metadata content on first selectDevice() failure, which is mostly like a transient failure. So let's dump metadata only on the last selectDevice() failure. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-29 21:27:33 +00:00