Shuaiyi Zhang

Does the storageclass have the annotation which can override global filesystemOverhead?

5

Refer to [CDI Configuration](https://github.com/kubevirt/containerized-data-importer/blob/main/doc/cdi-config.md), configure cdi/cdi like this can set filesystemOverhead of storageClass local-path. ``` spec: config: filesystemOverhead: global: "0.055" storageClass: local-path: "0.0" ``` Does the storageclass have the annotation...

docs: update registry config guide

2

Update registry config guide to use CRI service plugin type added in https://github.com/containerd/containerd/pull/9681 ``` - // CRIImagePlugin implements a cri image service - CRIImagePlugin plugin.Type = "io.containerd.cri.image.v1" + // CRIServicePlugin...

optimize iterations output

1

## Type of change - [ ] Refactor - [ ] New feature - [ ] Bug fix - [x] Optimization - [ ] Documentation Update ## Description Use `examples/workloads/kubelet-density`，and...

bpfman: don't panic if DatabaseLockError occurred

2

Currently `bpfman` will panic when DatabaseLockError occurred. ``` thread 'main' panicked at /root/go/src/github.com/bpfman/bpfman/bpfman/src/lib.rs:486:14: Unable to open root database: DatabaseLockError note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace ```...

[bug] Multi node TP CUDA error: invalid device ordinal

2

**Describe the bug** - LMCache `0.3.1.dev34` - vLLM `0.9.2.dev12+g7c644ab6d.precompiled` - 2 nodes, 1GPU/per node - tensor-parallel-size 2, vllm crash ``` INFO 06-11 19:21:45 [ray_distributed_executor.py:128] Shutting down Ray distributed executor. If...

[bugfix] skip set cuda device

1

FILL IN THE PR DESCRIPTION HERE FIX #814 The `VllmWorker` process sets the CUDA device using `local_rank` in `init_device` during initialization. Therefore, we don't need to change the CUDA device...

[Bugfix] Load dict from disk in LocalDiskBackend.__init__

1

FIX #562 Try to load `LocalDiskBackend.dict` from disk in `LocalDiskBackend.__init__`. After this PR, restarts vLLM and ask the same question. vLLM log ``` DEBUG 04-30 19:12:28 [core.py:427] EngineCore loop active....