[None][feat] Mooncake store part 1: pool, CLI, and V2 scheduler preemption - #19235

Open

brb-nv wants to merge 17 commits into

NVIDIA:mainfrom

brb-nv:user/brb/mooncake-integration-part-1

Commits on Sep 22, 2026

[None][feat] Mooncake store part 1: pool, CLI, and V2 scheduler preemption
brb-nv
committed
6ed271d
[None][chore] Address comments from CodeRabbit
brb-nv
committed
b965d58
[None][fix] Address comments from Thor
brb-nv
committed
3230c5c
[None][chore] Address follow-up comments from CodeRabbit
brb-nv
committed
22c44e6
[None][fix] Address comments from zhaoyuanh-nvidia
brb-nv
committed
c4a90cc
[None][fix] Offer --telemetry/--no-telemetry on the mooncake commands
brb-nv
committed
98d3039
[None][fix] Hand a signalled mooncake command to the telemetry boundary
brb-nv
committed
329191e
[None][fix] Do not call a pending KV transfer a V2 scheduler deadlock
brb-nv
committed
e69a4a8
[None][fix] Reserve preempted KV pages for the request that freed them
brb-nv
committed
391c17d
[None][chore] Tighten comments on the review follow-ups
brb-nv
committed
911db0b
[None][chore] Shorten the mooncake CMake package cleanup
brb-nv
committed
b81cf64

Commits on Sep 23, 2026

[None][chore] Defer the Mooncake LlmArgs surface to the connector MR
brb-nv
committed
2192fd6
Address comments from nv-xtf
brb-nv
committed
3eb863f
Address comments from Coderabbit
brb-nv
committed
138e5e3
Logging for scheduler deadlock
brb-nv
committed
865948a
[None][chore] Share the mooncake client pin between the wheel and container
brb-nv
committed
1e19d2b

Commits on Sep 24, 2026

[None][fix] Drop the Mooncake import check from the container build
brb-nv
committed
4f1fa06