chore(release): bump version to 0.22.4

[xiaguan]

chore(release): bump version to 0.22.4

Bumps the Rust workspace, the pegaflow-llm Python package, the commitizen version, and the Cargo.lock workspace package versions from 0.22.3 to 0.22.4.

---

Release notes — 0.22.3 → 0.22.4

18 PRs landed on master since v0.22.3 (2026-05-15). Grouped below for release notes.

Highlights

• Disaggregated prefill/decode over RDMA push (feat(pd): RDMA push connector for disaggregated prefill/decode #297) — a brand-new vLLM v1 KV connector (PdConnector) plus a v2 RDMA transfer engine (pegaflow-transfer/src/v2). KV is pushed prefill→decode layer-by-layer via one-sided RDMA WRITE as each attention layer completes, overlapping transfer with the forward pass instead of pulling after prefill finishes (vLLM NIXL model). On H20 / Qwen3-8B the added TTFT is 2–4× lower than NIXL across 512–16k input lengths.
• Query leases replace query pinning (feat(connector): replace query pinning with leases #284, fix(connector): allow query leases across workers #288) — the query/load/release control path moved from pin refcounts to lease-backed ownership. Query results collapse to Loading/Ready only; Ready carries num_hit_blocks plus an opaque lease that transfers scheduler→worker and is released on cleanup/failure, with a TTL sweeper reclaiming abandoned leases.
• Save-only connector mode (feat(connector): add save-only pegaflow mode #300) — new pegaflow.mode config; save_only skips Pega query/load while still advancing save metadata, so an instance can populate the cache without serving reads.

Features

• feat(pd): RDMA push connector for disaggregated prefill/decode (feat(pd): RDMA push connector for disaggregated prefill/decode #297)
• feat(connector): add save-only pegaflow mode (feat(connector): add save-only pegaflow mode #300)
• feat(storage): sharded SSD cache — cache spread across multiple files, uring engine dispatches across shards, prefetch ready-blocks now ordered by requested keys (feat(storage): add sharded SSD cache support #299)
• feat(rdma): per-peer N QPs with WQE-level round-robin — new --qps-per-peer (default 2), round-robin at WQE level so one in-flight task saturates all QPs; handshake validates both sides agree on N (feat(rdma): per-peer N QPs with WQE-level round-robin #291)
• feat(metaserver): node lifecycle fencing — heartbeat-based node tracking with per-node UUID fencing, stale nodes hidden via --node-stale-secs (feat(metaserver): add node lifecycle fencing #285, closes feat(metaserver): add server heartbeat liveness and graceful Bye RPC #222)
• feat(connector): replace query pinning with leases (feat(connector): replace query pinning with leases #284)

Fixes

• fix(connector): preserve non-MLA KV layout registration so cross-layer layouts (e.g. GLM-4.7-FP8) register by block stride; limit logical/physical block splitting to MLA (fix(connector): preserve non-MLA kv layout registration #295, closes PegaKVConnector fails to initialize with cross-layer KV cache layout #294)
• fix(numa): allocate pinned pools on GPU-local NUMA nodes instead of the full CPU NUMA set, avoiding wasted capacity on CPU-only nodes (fix(numa): allocate pinned pools on GPU-local NUMA nodes #293)
• fix(connector): handle split physical KV blocks — group split physical rows into one logical block when FlashMLA uses smaller physical blocks (fix(connector): handle split physical kv blocks #292)
• fix(connector): allow a query lease to be consumed once per registered worker, fixing multi-worker query lease is unknown or expired (fix(connector): allow query leases across workers #288)
• fix(server): fail on invalid RDMA NICs — accept comma/space-separated --nics, reject empty names, propagate RDMA init failures instead of silently disabling P2P (fix(server): fail on invalid rdma nics #283, fixes pegaflow-server: --nics silently misparses comma-separated values #276)
• fix(connector): remove the unused scheduler pending-save request limit and save-drop accounting (fix(connector): remove scheduler save limit #282)
• fix(connector): demote cache_lookup_reuse log from INFO to DEBUG to stop log spam under cache pressure (fix(connector): demote cache_lookup_reuse log to debug #280)

Performance

• perf: CPU-path Criterion benchmarks + long-block save optimizations — e.g. query_prefetch_lease/32768 ~12.3 ms → ~6.1 ms, save_flush_unique/8192 ~21.3 ms → ~13.1 ms via reduced prefix-key cloning, ordered multi-layer save grouping, and RawBlock inline-segment allocation (perf: add cpu path benchmarks and optimize long-block saves #290)

Internal / refactor / tests

• refactor(metrics): centralize histogram buckets behind a build_buckets helper (refactor(metrics): use build_buckets helper for histogram buckets #298)
• refactor(core): make prefetch tasks terminal (Ready results carry RAM prefix blocks); default storage admission to no TinyLFU unless explicitly enabled (refactor(core): make prefetch task terminal #287)
• test(server): mock vLLM gRPC E2E harness covering save/query/load/release/session contracts (test(server): add mock vLLM RPC E2E coverage #289)
• chore: tune transfer duration histogram buckets toward long-tail visibility (chore: tune transfer duration buckets #281)

Notable behavior & config changes (upgrade notes)

• Query API: query results are now Loading/Ready only; Ready exposes num_hit_blocks + an opaque lease. Pin/unpin refcount semantics are gone (feat(connector): replace query pinning with leases #284, fix(connector): allow query leases across workers #288).
• Release RPC: unknown/expired leases now return FailedPrecondition instead of being silently accepted (test(server): add mock vLLM RPC E2E coverage #289).
• --nics: now rejects empty entries (e.g. mlx5_0,,mlx5_1) and fails startup on RDMA init errors rather than silently falling back to no-P2P (fix(server): fail on invalid rdma nics #283).
• New CLI flags: --qps-per-peer (default 2) (feat(rdma): per-peer N QPs with WQE-level round-robin #291), --node-stale-secs for metaserver (feat(metaserver): add node lifecycle fencing #285).
• New connector config: pegaflow.mode with read_write (default) / save_only (feat(connector): add save-only pegaflow mode #300).
• Storage admission: TinyLFU is now off unless explicitly enabled (refactor(core): make prefetch task terminal #287).

Full PR list

#298 refactor(metrics): use build_buckets helper for histogram buckets
#300 feat(connector): add save-only pegaflow mode
#299 feat(storage): add sharded SSD cache support
#297 feat(pd): RDMA push connector for disaggregated prefill/decode
#290 perf: add cpu path benchmarks and optimize long-block saves
#295 fix(connector): preserve non-MLA kv layout registration
#293 fix(numa): allocate pinned pools on GPU-local NUMA nodes
#292 fix(connector): handle split physical kv blocks
#291 feat(rdma): per-peer N QPs with WQE-level round-robin
#285 feat(metaserver): add node lifecycle fencing
#287 refactor(core): make prefetch task terminal
#289 test(server): add mock vLLM RPC E2E coverage
#288 fix(connector): allow query leases across workers
#284 feat(connector): replace query pinning with leases
#283 fix(server): fail on invalid rdma nics
#282 fix(connector): remove scheduler save limit
#281 chore: tune transfer duration buckets
#280 fix(connector): demote cache_lookup_reuse log to debug

[feifei-111]

LGTM