Files
seaweedfs/weed/filer/reader_at_test.go
T
930603eb74 S3: optionally serve remote-mounted objects from remote when the local read fails (#10837)
* feat(s3): serve from remote on local read failure

When a locally-cached chunk of a remote-mounted object becomes unreadable
(volume server down/restarting, or an evicted needle 404ing under
retry-backoff), fall back to serving the object from its mounted remote
instead of erroring. A bounded pre-flight probe makes a stuck volume trip
the timeout rather than stalling the request.

Gated by -localReadFallbackToRemote (default off) with
-localReadFallbackTimeout (2s default), so existing deployments are
unaffected until they opt in.

* fix(s3): register local-read-fallback flags for mini/server/filer

The mini, server and filer launchers build S3Options directly and only
populate the flag pointers they register. Without registering the two new
flags there, startS3Server dereferenced nil pointers and crashed at boot,
failing every integration suite that runs `weed mini`.

* fix(s3): treat a zero-byte probe read as unreadable

A read that returns no byte -- whether it reports io.EOF or no error at all
-- means the offset is not locally readable, so the probe must fall back to
the remote rather than proceeding to stream a truncated response. Only a
returned byte (including the object's final byte with a trailing io.EOF)
counts as readable.

* s3: finish a mid-stream local read failure from the remote mount

The pre-flight probe only proves the byte at the requested offset readable.
A multi-chunk object can still lose a later chunk after the 200/206 and its
Content-Length are committed, which truncated the body with no fallback.
Resume from the mounted remote at the byte the local copy stopped at, so the
response still carries the declared length. A short local read that surfaces
as a clean EOF is treated the same way instead of silently truncating.

* s3: fall back to the remote mount without a CLI switch

Serving a remote-mounted object from its authoritative remote is what the
read should have done all along -- the alternative is a 500 on an object the
cluster can still reach -- so make it the behavior instead of two new flags,
with the probe bounded by a constant.

* s3: trim the comments on the fallback path

* filer: report only the contiguous prefix when a parallel chunk read fails

The parallel branch of doReadAt fans the chunk reads straight into their own
windows of the output buffer, then sums every task's bytesRead. A middle chunk
failing while a later one succeeds therefore returned a length covering a hole
the reader never filled, handing the caller zeros in the middle of otherwise
valid data.

* s3: only splice the remote onto a local prefix while it is the cached generation

Eligibility establishes a size match, not byte identity: a remote key
overwritten with same-size content between the cache fill and the fallback
would have finished the response with bytes from a second generation, under
the first one's ETag. Stat the remote before resuming and keep the local
error when it no longer matches -- a truncated body is a visible failure,
a spliced one is not.

---------

Co-authored-by: Chris Lu <chrislusf@users.noreply.github.com>
Co-authored-by: Chris Lu <chris.lu@gmail.com>
2026-08-21 11:47:53 -07:00

260 lines
6.9 KiB
Go

package filer
import (
"bytes"
"context"
"errors"
"io"
"math"
"strconv"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
type mockChunkCache struct {
}
func (m *mockChunkCache) GetChunk(fileId string, minSize uint64) (data []byte) {
x, _ := strconv.Atoi(fileId)
data = make([]byte, minSize)
for i := 0; i < int(minSize); i++ {
data[i] = byte(x)
}
return data
}
func (m *mockChunkCache) ReadChunkAt(data []byte, fileId string, offset uint64) (n int, err error) {
x, _ := strconv.Atoi(fileId)
for i := 0; i < len(data); i++ {
data[i] = byte(x)
}
return len(data), nil
}
func (m *mockChunkCache) SetChunk(fileId string, data []byte) {
}
func (m *mockChunkCache) GetMaxFilePartSizeInCache() uint64 {
return 0
}
func (m *mockChunkCache) IsInCache(fileId string, lockNeeded bool) (answer bool) {
return false
}
func TestReaderAt(t *testing.T) {
visibles := NewIntervalList[*VisibleInterval]()
addVisibleInterval(visibles, &VisibleInterval{
start: 1,
stop: 2,
fileId: "1",
chunkSize: 9,
})
addVisibleInterval(visibles, &VisibleInterval{
start: 3,
stop: 4,
fileId: "3",
chunkSize: 1,
})
addVisibleInterval(visibles, &VisibleInterval{
start: 5,
stop: 6,
fileId: "5",
chunkSize: 2,
})
addVisibleInterval(visibles, &VisibleInterval{
start: 7,
stop: 9,
fileId: "7",
chunkSize: 2,
})
addVisibleInterval(visibles, &VisibleInterval{
start: 9,
stop: 10,
fileId: "9",
chunkSize: 2,
})
readerAt := &ChunkReadAt{
chunkViews: ViewFromVisibleIntervals(visibles, 0, math.MaxInt64),
fileSize: 10,
readerCache: NewReaderCache(3, &mockChunkCache{}, nil, nil),
readerPattern: NewReaderPattern(),
}
testReadAt(t, readerAt, 0, 10, 10, io.EOF, nil, nil)
testReadAt(t, readerAt, 0, 12, 10, io.EOF, nil, nil)
testReadAt(t, readerAt, 2, 8, 8, io.EOF, nil, nil)
testReadAt(t, readerAt, 3, 6, 6, nil, nil, nil)
}
func testReadAt(t *testing.T, readerAt *ChunkReadAt, offset int64, size int, expectedN int, expectedErr error, data, expectedData []byte) {
if data == nil {
data = make([]byte, size)
}
n, _, err := readerAt.doReadAt(context.Background(), data, offset)
if expectedN != n {
t.Errorf("unexpected read size: %d, expect: %d", n, expectedN)
}
if err != expectedErr {
t.Errorf("unexpected read error: %v, expect: %v", err, expectedErr)
}
if expectedData != nil && !bytes.Equal(data, expectedData) {
t.Errorf("unexpected read data: %v, expect: %v", data, expectedData)
}
}
func TestReaderAt0(t *testing.T) {
visibles := NewIntervalList[*VisibleInterval]()
addVisibleInterval(visibles, &VisibleInterval{
start: 2,
stop: 5,
fileId: "1",
chunkSize: 9,
})
addVisibleInterval(visibles, &VisibleInterval{
start: 7,
stop: 9,
fileId: "2",
chunkSize: 9,
})
readerAt := &ChunkReadAt{
chunkViews: ViewFromVisibleIntervals(visibles, 0, math.MaxInt64),
fileSize: 10,
readerCache: NewReaderCache(3, &mockChunkCache{}, nil, nil),
readerPattern: NewReaderPattern(),
}
testReadAt(t, readerAt, 0, 10, 10, io.EOF, nil, nil)
testReadAt(t, readerAt, 3, 16, 7, io.EOF, nil, nil)
testReadAt(t, readerAt, 3, 5, 5, nil, nil, nil)
testReadAt(t, readerAt, 11, 5, 0, io.EOF, nil, nil)
testReadAt(t, readerAt, 10, 5, 0, io.EOF, nil, nil)
}
func TestReaderAt1(t *testing.T) {
visibles := NewIntervalList[*VisibleInterval]()
addVisibleInterval(visibles, &VisibleInterval{
start: 2,
stop: 5,
fileId: "1",
chunkSize: 9,
})
readerAt := &ChunkReadAt{
chunkViews: ViewFromVisibleIntervals(visibles, 0, math.MaxInt64),
fileSize: 20,
readerCache: NewReaderCache(3, &mockChunkCache{}, nil, nil),
readerPattern: NewReaderPattern(),
}
testReadAt(t, readerAt, 0, 20, 20, io.EOF, nil, nil)
testReadAt(t, readerAt, 1, 7, 7, nil, nil, nil)
testReadAt(t, readerAt, 0, 1, 1, nil, nil, nil)
testReadAt(t, readerAt, 18, 4, 2, io.EOF, nil, nil)
testReadAt(t, readerAt, 12, 4, 4, nil, nil, nil)
testReadAt(t, readerAt, 4, 20, 16, io.EOF, nil, nil)
testReadAt(t, readerAt, 4, 10, 10, nil, nil, nil)
testReadAt(t, readerAt, 1, 10, 10, nil, nil, nil)
}
func TestReaderAtGappedChunksDoNotLeak(t *testing.T) {
visibles := NewIntervalList[*VisibleInterval]()
addVisibleInterval(visibles, &VisibleInterval{
start: 2,
stop: 3,
fileId: "1",
chunkSize: 5,
})
addVisibleInterval(visibles, &VisibleInterval{
start: 7,
stop: 9,
fileId: "1",
chunkSize: 4,
})
readerAt := &ChunkReadAt{
chunkViews: ViewFromVisibleIntervals(visibles, 0, math.MaxInt64),
fileSize: 9,
readerCache: NewReaderCache(3, &mockChunkCache{}, nil, nil),
readerPattern: NewReaderPattern(),
}
testReadAt(t, readerAt, 0, 9, 9, io.EOF, []byte{2, 2, 2, 2, 2, 2, 2, 2, 2}, []byte{0, 0, 1, 0, 0, 0, 0, 1, 1})
testReadAt(t, readerAt, 1, 8, 8, io.EOF, []byte{2, 2, 2, 2, 2, 2, 2, 2}, []byte{0, 1, 0, 0, 0, 0, 1, 1})
}
func TestReaderAtSparseFileDoesNotLeak(t *testing.T) {
readerAt := &ChunkReadAt{
chunkViews: ViewFromVisibleIntervals(NewIntervalList[*VisibleInterval](), 0, math.MaxInt64),
fileSize: 3,
readerCache: NewReaderCache(3, &mockChunkCache{}, nil, nil),
readerPattern: NewReaderPattern(),
}
testReadAt(t, readerAt, 0, 3, 3, io.EOF, []byte{2, 2, 2}, []byte{0, 0, 0})
testReadAt(t, readerAt, 1, 2, 2, io.EOF, []byte{2, 2}, []byte{0, 0})
}
// holeChunkCache serves every chunk but one, so that chunk falls through to a
// lookup that fails.
type holeChunkCache struct {
mockChunkCache
missingFileId string
}
func (c *holeChunkCache) ReadChunkAt(data []byte, fileId string, offset uint64) (int, error) {
if fileId == c.missingFileId {
return 0, nil
}
return c.mockChunkCache.ReadChunkAt(data, fileId, offset)
}
func (c *holeChunkCache) GetMaxFilePartSizeInCache() uint64 {
return math.MaxUint64
}
// The parallel reads place their bytes directly in the output buffer, so a
// failing middle chunk leaves a hole with valid data after it. Reporting that
// length would hand the caller bytes it never read.
func TestReaderAtParallelFailureReturnsContiguousPrefix(t *testing.T) {
visibles := NewIntervalList[*VisibleInterval]()
for i, fileId := range []string{"1", "2", "3"} {
addVisibleInterval(visibles, &VisibleInterval{
start: int64(i) * 4,
stop: int64(i)*4 + 4,
fileId: fileId,
chunkSize: 4,
})
}
readerAt := &ChunkReadAt{
ctx: context.Background(),
chunkViews: ViewFromVisibleIntervals(visibles, 0, math.MaxInt64),
fileSize: 12,
readerCache: NewReaderCache(3, &holeChunkCache{missingFileId: "2"}, func(ctx context.Context, fileId string) ([]string, error) {
return nil, errors.New("volume down")
}, nil),
readerPattern: NewReaderPattern(),
prefetchCount: 3,
}
buf := make([]byte, 12)
n, err := readerAt.ReadAt(buf, 0)
require.Error(t, err)
assert.Equal(t, 4, n, "only the bytes before the failed chunk are readable")
assert.Equal(t, []byte{1, 1, 1, 1}, buf[:n])
}