Optimize Rust CI builds with proper sccache configuration (#15581)

.github/workflows/pr-benchmark-rust.yml

@@ -40,7 +40,7 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2
@@ -87,7 +87,7 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2
@@ -159,7 +159,7 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2
@@ -226,7 +226,7 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2

.github/workflows/pr-test-pd-router.yml

@@ -20,6 +20,10 @@ concurrency:
   group: test-disaggregation-${{ github.ref }}
   cancel-in-progress: true
 
+env:
+  RUSTC_WRAPPER: sccache
+  SCCACHE_GHA_ENABLED: "true"
+
 permissions:
   contents: read
   pull-requests: write
@@ -50,18 +54,19 @@ jobs:
       run: |
         bash scripts/ci/ci_install_rust.sh
 
-    - name: Cache Rust dependencies
-      uses: actions/cache@v4
+    - name: Configure sccache
+      uses: mozilla-actions/sccache-action@v0.0.9
       with:
-        path: |
-          ~/.cargo/bin/
-          ~/.cargo/registry/index/
-          ~/.cargo/registry/cache/
-          ~/.cargo/git/db/
-          sgl-model-gateway/target/
-        key: ${{ runner.os }}-cargo-${{ hashFiles('sgl-model-gateway/Cargo.lock') }}
-        restore-keys: |
-          ${{ runner.os }}-cargo-
+        version: "v0.12.0"
+
+    - name: Rust cache
+      uses: Swatinem/rust-cache@v2
+      with:
+        workspaces: sgl-model-gateway
+        shared-key: "rust-cache"
+        cache-all-crates: true
+        cache-on-failure: true
+        save-if: true
 
     - name: Cache pip dependencies
       uses: actions/cache@v4

.github/workflows/pr-test-rust.yml

@@ -55,7 +55,16 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
+
+      - name: Rust cache
+        uses: Swatinem/rust-cache@v2
+        with:
+          workspaces: "."
+          shared-key: "rust-cache"
+          cache-all-crates: true
+          cache-on-failure: true
+          save-if: true
 
       - name: Test maturin build
         uses: PyO3/maturin-action@v1
@@ -91,14 +100,16 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2
         with:
           workspaces: sgl-model-gateway
+          shared-key: "rust-cache"
           cache-all-crates: true
           cache-on-failure: true
+          save-if: true
 
       - name: Run lint
         run: |
@@ -153,14 +164,16 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2
         with:
           workspaces: sgl-model-gateway
+          shared-key: "rust-cache"
           cache-all-crates: true
           cache-on-failure: true
+          save-if: true
 
       - name: Install SGLang dependencies
         run: |
@@ -225,14 +238,16 @@ jobs:
       - name: Configure sccache
         uses: mozilla-actions/sccache-action@v0.0.9
         with:
-          version: "v0.10.0"
+          version: "v0.12.0"
 
       - name: Rust cache
         uses: Swatinem/rust-cache@v2
         with:
           workspaces: sgl-model-gateway
+          shared-key: "rust-cache"
           cache-all-crates: true
           cache-on-failure: true
+          save-if: true
 
       - name: Install SGLang dependencies
         run: |

sgl-model-gateway/README.md

@@ -625,6 +625,27 @@ pytest py_test/
 ```
 For production builds, use `maturin build --release --out dist` from the `bindings/python/` directory to create optimized wheels. During development, `maturin develop` rebuilds and installs instantly without creating wheel files. Use `python -m sglang_router.launch_server` to co-launch router and SGLang workers in small clusters for local validation.
 
+### Build Caching
+
+**Local development** uses incremental compilation by default (configured in `.cargo/config.toml`), which is optimal for the edit-compile-test cycle.
+
+**For release builds or CI**, you can optionally use [sccache](https://github.com/mozilla/sccache) to cache compilation artifacts:
+
+```bash
+# Install sccache
+cargo install sccache
+
+# Option 1: Set environment variable (per-session)
+export RUSTC_WRAPPER=sccache
+cargo build --release
+
+# Option 2: Add to your global cargo config (~/.cargo/config.toml)
+# [build]
+# rustc-wrapper = "sccache"
+```
+
+> **Note:** sccache and incremental compilation are mutually exclusive—sccache cannot cache incrementally compiled crates. The project defaults to incremental compilation for faster local iteration. Use sccache for clean/release builds where caching across builds matters more.
+
 ---
 
 ## Release Management