Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix (almost) infinite loop in Fileset writer when previous fileset encountered an error writing out index files #2058

Merged
merged 4 commits into from
Dec 5, 2019
Merged
Show file tree
Hide file tree
Changes from 3 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions src/dbnode/persist/fs/read_write_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -494,12 +494,6 @@ func TestWriterOnlyWritesNonNilBytes(t *testing.T) {
filePathPrefix := filepath.Join(dir, "")
defer os.RemoveAll(dir)

checkedBytes := func(b []byte) checked.Bytes {
r := checked.NewBytes(b, nil)
r.IncRef()
return r
}

w := newTestWriter(t, filePathPrefix)
writerOpts := DataWriterOpenOptions{
BlockSize: testBlockSize,
Expand All @@ -526,3 +520,9 @@ func TestWriterOnlyWritesNonNilBytes(t *testing.T) {
{"foo", nil, []byte{1, 2, 3, 4, 5, 6}},
})
}

func checkedBytes(b []byte) checked.Bytes {
r := checked.NewBytes(b, nil)
r.IncRef()
return r
}
5 changes: 5 additions & 0 deletions src/dbnode/persist/fs/write.go
Original file line number Diff line number Diff line change
Expand Up @@ -161,6 +161,11 @@ func (w *writer) Open(opts DataWriterOpenOptions) error {
w.currIdx = 0
w.currOffset = 0
w.err = nil
// This happens after writing the previous set of files index files, however, do it
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

mind grouping the reseting code into a new func (w *writer) reset() fn

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

tangentially - should indexEntries be bounded/randomly re-allocated if it exceeds certain size to reduce memory usage? not for this PR, just curious myself.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@prateek sure

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@vdarulis Theoretically yeah it probably should. In practice we only keep one of these around per node (for the most part) and the size of this slice will never exceed the number of series for a given block/shard combination so in practice its usually in the 10s of thousands tops and each item in the slice is not that big so I think its really unlikely it would become an issue. There are definitely other things like this in the code-base though where we're more paranoid because they could become an issue if left unchecked

// again to ensure they get cleared even if there was a premature error writing out the
// previous set of files which would have prevented them from being cleared.
w.indexEntries.releaseRefs()
w.indexEntries = w.indexEntries[:0]

var (
shardDir string
Expand Down
67 changes: 67 additions & 0 deletions src/dbnode/persist/fs/write_test.go
Original file line number Diff line number Diff line change
@@ -0,0 +1,67 @@
// Copyright (c) 2019 Uber Technologies, Inc.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.

package fs

import (
"os"
"path/filepath"
"testing"
"time"

"github.com/m3db/m3/src/dbnode/persist"
"github.com/m3db/m3/src/x/ident"
"github.com/stretchr/testify/require"
)

// TestWriteReuseAfterError was added as a regression test after it was
// discovered that reusing a fileset writer after a called to Close() had
// returned an error could make the fileset writer end up in a near infinite
// loop when it was reused to write out a completely indepedent set of files.
//
// This test verifies that the fix works as expected and prevents regressions
// of the issue.
func TestWriteReuseAfterError(t *testing.T) {
dir := createTempDir(t)
filePathPrefix := filepath.Join(dir, "")
defer os.RemoveAll(dir)

seriesID := ident.StringID("series1")
w := newTestWriter(t, filePathPrefix)
writerOpts := DataWriterOpenOptions{
Identifier: FileSetFileIdentifier{
Namespace: testNs1ID,
Shard: 0,
BlockStart: time.Now().Truncate(time.Hour),
VolumeIndex: 0,
},
BlockSize: time.Hour,
FileSetType: persist.FileSetFlushType,
}
data := checkedBytes([]byte{1, 2, 3})

require.NoError(t, w.Open(writerOpts))
require.NoError(t, w.Write(seriesID, ident.Tags{}, data, 0))
require.NoError(t, w.Write(seriesID, ident.Tags{}, data, 0))
require.Error(t, w.Close())

require.NoError(t, w.Open(writerOpts))
require.NoError(t, w.Close())
}