make static cache policy explicit, disable caching for objects >100 MiB

syphar · syphar · commit 4db250524e25 · 2025-12-05T14:01:54.000+01:00
diff --git a/src/web/cache.rs b/src/web/cache.rs
@@ -18,6 +18,9 @@ use tracing::error;
 /// This enables us to use the fastly "soft purge" for everything.
 pub const SURROGATE_KEY_ALL: SurrogateKey = SurrogateKey::from_static("all");
 
+/// cache poicy for static assets like rustdoc files or build assets.
+pub const STATIC_ASSET_CACHE_POLICY: CachePolicy = CachePolicy::ForeverInCdnAndBrowser;
+
 #[derive(Debug, Clone, PartialEq)]
 pub struct ResponseCacheHeaders {
     pub cache_control: Option<HeaderValue>,
diff --git a/src/web/file.rs b/src/web/file.rs
@@ -18,6 +18,12 @@ use axum_extra::{
 };
 use std::time::SystemTime;
 use tokio_util::io::ReaderStream;
+use tracing::warn;
+
+// https://docs.fastly.com/products/compute-resource-limits#default-limits
+// https://www.fastly.com/documentation/guides/full-site-delivery/performance/failure-modes-with-large-objects/
+// https://www.fastly.com/documentation/guides/full-site-delivery/caching/segmented-caching/
+const FASTLY_CACHE_MAX_OBJECT_SIZE: usize = 100 * 1024 * 1024; // 100 MB
 
 #[derive(Debug)]
 pub(crate) struct File(pub(crate) Blob);
@@ -41,9 +47,13 @@ impl File {
 
 #[cfg(test)]
 impl File {
-    pub fn into_response(self, if_none_match: Option<&IfNoneMatch>) -> AxumResponse {
+    pub fn into_response(
+        self,
+        if_none_match: Option<&IfNoneMatch>,
+        cache_policy: CachePolicy,
+    ) -> AxumResponse {
         let streaming_blob: StreamingBlob = self.0.into();
-        StreamingFile(streaming_blob).into_response(if_none_match)
+        StreamingFile(streaming_blob).into_response(if_none_match, cache_policy)
     }
 }
 
@@ -56,8 +66,40 @@ impl StreamingFile {
         Ok(StreamingFile(storage.get_stream(path).await?))
     }
 
-    pub fn into_response(self, if_none_match: Option<&IfNoneMatch>) -> AxumResponse {
-        const CACHE_POLICY: CachePolicy = CachePolicy::ForeverInCdnAndBrowser;
+    pub fn into_response(
+        self,
+        if_none_match: Option<&IfNoneMatch>,
+        mut cache_policy: CachePolicy,
+    ) -> AxumResponse {
+        // by default Fastly can only cache objects up to 100 MiB.
+        // Since we're streaming the response via chunked encoding, fastly itself doesn't know
+        // the object size until the streamed data size is > 100 MiB. In this case fastly just
+        // cuts the connection.
+        // To avoid issues with caching large files, we disable CDN caching for files that are too
+        // big.
+        //
+        // See:
+        //   https://docs.fastly.com/products/compute-resource-limits#default-limits
+        //   https://www.fastly.com/documentation/guides/full-site-delivery/performance/failure-modes-with-large-objects/
+        //   https://www.fastly.com/documentation/guides/full-site-delivery/caching/segmented-caching/
+        //
+        // For now I use the `NoStoreMustRevalidate` policy, the important cache-control statement
+        // is only the `no-store` part.
+        //
+        // Future optimization could be:
+        // * only forbid fastly to storstore, and browsers still could.
+        // * implement segmented caching for large files somehow.
+        if self.0.content_length > FASTLY_CACHE_MAX_OBJECT_SIZE
+            && !matches!(cache_policy, CachePolicy::NoStoreMustRevalidate)
+        {
+            warn!(
+                storage_path = self.0.path,
+                content_length = self.0.content_length,
+                "Disabling CDN caching for large file"
+            );
+            cache_policy = CachePolicy::NoStoreMustRevalidate;
+        }
+
         let last_modified = LastModified::from(SystemTime::from(self.0.date_updated));
 
         if let Some(if_none_match) = if_none_match
@@ -69,7 +111,7 @@ impl StreamingFile {
                 // it's generally recommended to repeat caching headers on 304 responses
                 TypedHeader(etag.clone()),
                 TypedHeader(last_modified),
-                Extension(CACHE_POLICY),
+                Extension(cache_policy),
             )
                 .into_response()
         } else {
@@ -81,7 +123,7 @@ impl StreamingFile {
                 TypedHeader(ContentType::from(self.0.mime)),
                 TypedHeader(last_modified),
                 self.0.etag.map(TypedHeader),
-                Extension(CACHE_POLICY),
+                Extension(cache_policy),
                 Body::from_stream(stream),
             )
                 .into_response()
@@ -92,12 +134,18 @@ impl StreamingFile {
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::{storage::CompressionAlgorithm, test::TestEnvironment, web::headers::compute_etag};
+    use crate::{
+        storage::CompressionAlgorithm,
+        test::TestEnvironment,
+        web::{cache::STATIC_ASSET_CACHE_POLICY, headers::compute_etag},
+    };
     use axum_extra::headers::{ETag, HeaderMapExt as _};
     use chrono::Utc;
     use http::header::{CACHE_CONTROL, ETAG, LAST_MODIFIED};
     use std::{io, rc::Rc};
 
+    const CONTENT: &[u8] = b"Hello, world!";
+
     fn streaming_blob(
         content: impl Into<Vec<u8>>,
         alg: Option<CompressionAlgorithm>,
@@ -114,13 +162,28 @@ mod tests {
         }
     }
 
+    #[test]
+    fn test_big_file_stream_drops_cache_policy() {
+        let mut stream = streaming_blob(CONTENT, None);
+        stream.content_length = FASTLY_CACHE_MAX_OBJECT_SIZE + 1;
+
+        let response =
+            StreamingFile(stream).into_response(None, CachePolicy::ForeverInCdnAndBrowser);
+        // even though we passed a cache policy in `into_response`, it should be overridden to
+        // `NoCaching` due to the large size of the file.
+        let cache = response
+            .extensions()
+            .get::<CachePolicy>()
+            .expect("missing cache response extension");
+        assert!(matches!(cache, CachePolicy::NoStoreMustRevalidate));
+    }
+
     #[tokio::test]
     async fn test_stream_into_response() -> Result<()> {
-        const CONTENT: &[u8] = b"Hello, world!";
         let etag: ETag = {
             // first request normal
             let stream = StreamingFile(streaming_blob(CONTENT, None));
-            let resp = stream.into_response(None);
+            let resp = stream.into_response(None, STATIC_ASSET_CACHE_POLICY);
             assert!(resp.status().is_success());
             assert!(resp.headers().get(CACHE_CONTROL).is_none());
             let cache = resp
@@ -138,7 +201,7 @@ mod tests {
         {
             // cached request
             let stream = StreamingFile(streaming_blob(CONTENT, None));
-            let resp = stream.into_response(Some(&if_none_match));
+            let resp = stream.into_response(Some(&if_none_match), STATIC_ASSET_CACHE_POLICY);
             assert_eq!(resp.status(), StatusCode::NOT_MODIFIED);
 
             // cache related headers are repeated on the not-modified response
@@ -172,7 +235,7 @@ mod tests {
 
         file.0.date_updated = now;
 
-        let resp = file.into_response(None);
+        let resp = file.into_response(None, STATIC_ASSET_CACHE_POLICY);
         assert!(resp.status().is_success());
         assert!(resp.headers().get(CACHE_CONTROL).is_none());
         let cache = resp
diff --git a/src/web/rustdoc.rs b/src/web/rustdoc.rs
@@ -11,7 +11,7 @@ use crate::{
     utils::{self, Dependency},
     web::{
         MetaData, ReqVersion, axum_cached_redirect,
-        cache::CachePolicy,
+        cache::{CachePolicy, STATIC_ASSET_CACHE_POLICY},
         crate_details::CrateDetails,
         csp::Csp,
         error::{AxumNope, AxumResult},
@@ -228,7 +228,7 @@ async fn try_serve_legacy_toolchain_asset(
     // which is reached via the new handler.
     Ok(StreamingFile::from_path(&storage, &path)
         .await?
-        .into_response(if_none_match))
+        .into_response(if_none_match, STATIC_ASSET_CACHE_POLICY))
 }
 
 /// Handler called for `/:crate` and `/:crate/:version` URLs. Automatically redirects to the docs
@@ -344,7 +344,8 @@ pub(crate) async fn rustdoc_redirector_handler(
                 )
                 .await
             {
-                Ok(blob) => Ok(StreamingFile(blob).into_response(if_none_match.as_deref())),
+                Ok(blob) => Ok(StreamingFile(blob)
+                    .into_response(if_none_match.as_deref(), STATIC_ASSET_CACHE_POLICY)),
                 Err(err) => {
                     if !matches!(err.downcast_ref(), Some(AxumNope::ResourceNotFound))
                         && !matches!(err.downcast_ref(), Some(crate::storage::PathNotFoundError))
@@ -763,7 +764,9 @@ pub(crate) async fn rustdoc_html_server_handler(
         // default asset caching behaviour is `Cache::ForeverInCdnAndBrowser`.
         // This is an edge-case when we serve invocation specific static assets under `/latest/`:
         // https://github.com/rust-lang/docs.rs/issues/1593
-        return Ok(StreamingFile(blob).into_response(if_none_match.as_deref()));
+        return Ok(
+            StreamingFile(blob).into_response(if_none_match.as_deref(), STATIC_ASSET_CACHE_POLICY)
+        );
     }
 
     let latest_release = krate.latest_release()?;
@@ -982,9 +985,11 @@ pub(crate) async fn json_download_handler(
         Some(wanted_compression),
     );
 
+    let cache_policy = CachePolicy::ForeverInCdn(krate.name.clone().into());
+
     let (mut response, updated_storage_path) = match storage.get_raw_stream(&storage_path).await {
         Ok(file) => (
-            StreamingFile(file).into_response(if_none_match.as_deref()),
+            StreamingFile(file).into_response(if_none_match.as_deref(), cache_policy),
             None,
         ),
         Err(err) if matches!(err.downcast_ref(), Some(crate::storage::PathNotFoundError)) => {
@@ -1003,7 +1008,7 @@ pub(crate) async fn json_download_handler(
                 // redirect to that as fallback
                 (
                     StreamingFile(storage.get_raw_stream(&storage_path).await?)
-                        .into_response(if_none_match.as_deref()),
+                        .into_response(if_none_match.as_deref(), cache_policy),
                     Some(storage_path),
                 )
             } else {
@@ -1013,13 +1018,6 @@ pub(crate) async fn json_download_handler(
         Err(err) => return Err(err.into()),
     };
 
-    // StreamingFile::into_response automatically set the default cache-policy for
-    // static assets (ForeverInCdnAndBrowser).
-    // Here we override it with the standard policy for build output.
-    response
-        .extensions_mut()
-        .insert(CachePolicy::ForeverInCdn(krate.name.clone().into()));
-
     // set content-disposition to attachment to trigger download in browsers
     // For the attachment filename we can use just the filename without the path,
     // since that already contains all the info.
@@ -1059,15 +1057,10 @@ pub(crate) async fn download_handler(
     let version = &matched_release.release.version;
     let archive_path = rustdoc_archive_path(params.name(), version);
 
-    let mut response = StreamingFile(storage.get_raw_stream(&archive_path).await?)
-        .into_response(if_none_match.as_deref());
-
-    // StreamingFile::into_response automatically set the default cache-policy for
-    // static assets (ForeverInCdnAndBrowser).
-    // Here we override it with the standard policy for build output.
-    response
-        .extensions_mut()
-        .insert(CachePolicy::ForeverInCdn(matched_release.name.into()));
+    let mut response = StreamingFile(storage.get_raw_stream(&archive_path).await?).into_response(
+        if_none_match.as_deref(),
+        CachePolicy::ForeverInCdn(matched_release.name.into()),
+    );
 
     // set content-disposition to attachment to trigger download in browsers
     response.headers_mut().insert(
@@ -1092,7 +1085,7 @@ pub(crate) async fn static_asset_handler(
 
     Ok(StreamingFile::from_path(&storage, &storage_path)
         .await?
-        .into_response(if_none_match.as_deref()))
+        .into_response(if_none_match.as_deref(), STATIC_ASSET_CACHE_POLICY))
 }
 
 #[cfg(test)]
diff --git a/src/web/source.rs b/src/web/source.rs
@@ -5,15 +5,14 @@ use crate::{
     storage::PathNotFoundError,
     web::{
         MetaData, ReqVersion,
-        cache::CachePolicy,
+        cache::{CachePolicy, STATIC_ASSET_CACHE_POLICY},
         error::{AxumNope, AxumResult},
         extractors::{
             DbConnection,
             rustdoc::{PageKind, RustdocParams},
         },
         file::StreamingFile,
-        headers::CanonicalUrl,
-        headers::IfNoneMatch,
+        headers::{CanonicalUrl, IfNoneMatch},
         match_version,
         page::templates::{RenderBrands, RenderRegular, RenderSolid, filters},
     },
@@ -285,7 +284,8 @@ pub(crate) async fn source_browser_handler(
         let is_text = stream.mime.type_() == mime::TEXT || stream.mime == mime::APPLICATION_JSON;
         if !is_text {
             // if the file isn't text, serve it directly to the client
-            let mut response = StreamingFile(stream).into_response(if_none_match.as_deref());
+            let mut response = StreamingFile(stream)
+                .into_response(if_none_match.as_deref(), STATIC_ASSET_CACHE_POLICY);
             response.headers_mut().typed_insert(canonical_url);
             response
                 .extensions_mut()
diff --git a/src/web/statics.rs b/src/web/statics.rs
@@ -1,7 +1,7 @@
 use super::{
     cache::CachePolicy, headers::IfNoneMatch, metrics::request_recorder, routes::get_static,
 };
-use crate::db::mimes::APPLICATION_OPENSEARCH_XML;
+use crate::{db::mimes::APPLICATION_OPENSEARCH_XML, web::cache::STATIC_ASSET_CACHE_POLICY};
 use axum::{
     Router as AxumRouter,
     extract::{Extension, Request},
@@ -24,13 +24,11 @@ const RUSTDOC_2021_12_05_CSS: &str =
 const RUSTDOC_2025_08_20_CSS: &str =
     include_str!(concat!(env!("OUT_DIR"), "/rustdoc-2025-08-20.css"));
 
-const STATIC_CACHE_POLICY: CachePolicy = CachePolicy::ForeverInCdnAndBrowser;
-
 include!(concat!(env!("OUT_DIR"), "/static_etag_map.rs"));
 
 fn build_static_css_response(content: &'static str) -> impl IntoResponse {
     (
-        Extension(STATIC_CACHE_POLICY),
+        Extension(STATIC_ASSET_CACHE_POLICY),
         TypedHeader(ContentType::from(mime::TEXT_CSS)),
         content,
     )
@@ -43,7 +41,7 @@ async fn set_needed_static_headers(req: Request, next: Next) -> Response {
     let mut response = next.run(req).await;
 
     if response.status().is_success() {
-        response.extensions_mut().insert(STATIC_CACHE_POLICY);
+        response.extensions_mut().insert(STATIC_ASSET_CACHE_POLICY);
     }
 
     if is_opensearch_xml {
