lib/s3util: stream retrieved S3 data to disk on Get (#11)

On Get, forward the reader from the S3 transaction through to the disk store to
avoid pulling the fetched value into memory in the plugin.

Author: mikeauclair

lib/gobuild/gobuild.go

@@ -6,7 +6,6 @@
 package gobuild
 
 import (
-	"bytes"
 	"context"
 	"errors"
 	"expvar"
@@ -121,21 +120,22 @@ func (s *S3Cache) Get(ctx context.Context, actionID string) (outputID, diskPath
 		return "", "", err
 	}
 
-	object, err := s.S3Client.GetData(ctx, s.outputKey(outputID))
+	object, size, err := s.S3Client.Get(ctx, s.outputKey(outputID))
 	if err != nil {
 		// At this point we know the action exists, so if we can't read the
 		// object report it as an error rather than a cache miss.
 		return "", "", fmt.Errorf("[s3] read object %s: %w", outputID, err)
 	}
+	defer object.Close()
 	s.getFaultHit.Add(1)
 
 	// Now we should have the body; poke it into the local cache.  Preserve the
 	// modification timestamp recorded with the original action.
 	diskPath, err = s.Local.Put(ctx, gocache.Object{
 		ActionID: actionID,
 		OutputID: outputID,
-		Size:     int64(len(object)),
-		Body:     bytes.NewReader(object),
+		Size:     size,
+		Body:     object,
 		ModTime:  mtime,
 	})
 	return outputID, diskPath, err

lib/modproxy/modproxy.go

@@ -158,7 +158,7 @@ func (c *S3Cacher) Get(ctx context.Context, name string) (_ io.ReadCloser, oerr
 	}
 	defer c.sema.Release(1)
 
-	obj, err := c.S3Client.Get(ctx, c.makeKey(hash))
+	obj, _, err := c.S3Client.Get(ctx, c.makeKey(hash))
 	if errors.Is(err, fs.ErrNotExist) {
 		c.getFaultMiss.Add(1)
 		return nil, err

lib/s3util/s3util.go

@@ -122,24 +122,24 @@ func (c *Client) Put(ctx context.Context, key string, data io.Reader) error {
 // close the reader when finished.
 //
 // If the key is not found, the resulting error satisfies [fs.ErrNotExist].
-func (c *Client) Get(ctx context.Context, key string) (io.ReadCloser, error) {
+func (c *Client) Get(ctx context.Context, key string) (io.ReadCloser, int64, error) {
 	rsp, err := c.Client.GetObject(ctx, &s3.GetObjectInput{
 		Bucket: &c.Bucket,
 		Key:    &key,
 	})
 	if err != nil {
 		if IsNotExist(err) {
-			return nil, fmt.Errorf("key %q: %w", key, fs.ErrNotExist)
+			return nil, -1, fmt.Errorf("key %q: %w", key, fs.ErrNotExist)
 		}
-		return nil, err
+		return nil, -1, err
 	}
-	return rsp.Body, nil
+	return rsp.Body, *rsp.ContentLength, nil
 }
 
 // GetData returns the contents of the specified key from S3. It is a shorthand
 // for calling Get followed by io.ReadAll on the result.
 func (c *Client) GetData(ctx context.Context, key string) ([]byte, error) {
-	rc, err := c.Get(ctx, key)
+	rc, _, err := c.Get(ctx, key)
 	if err != nil {
 		return nil, err
 	}