Copilot commented on code in PR #670:
URL:
https://github.com/apache/skywalking-banyandb/pull/670#discussion_r2094160197
##########
banyand/measure/merger.go:
##########
@@ -260,7 +260,10 @@ func mergeParts(fileSystem fs.FileSystem, closeCh <-chan
struct{}, parts []*part
}
pm.mustWriteMetadata(fileSystem, dstPath)
fileSystem.SyncPath(dstPath)
- p := mustOpenFilePart(partID, root, fileSystem)
+ for i := range parts {
+ cache.delete(parts[i].ID())
+ }
+ p := mustOpenFilePart(partID, root, fileSystem, nil)
Review Comment:
In `mergeParts`, the merged part is opened with `nil` cache even though a
`cache` parameter is provided. This prevents caching of metadata on merged
parts. It should pass the `cache` argument instead of `nil`.
```suggestion
p := mustOpenFilePart(partID, root, fileSystem, cache)
```
##########
banyand/measure/flusher.go:
##########
@@ -153,7 +153,7 @@ func (tst *tsTable) flush(snapshot *snapshot, flushCh chan
*flusherIntroduction)
partsCount++
partPath := partPath(tst.root, pw.ID())
pw.mp.mustFlush(tst.fileSystem, partPath)
- newPW := newPartWrapper(nil, mustOpenFilePart(pw.ID(),
tst.root, tst.fileSystem))
+ newPW := newPartWrapper(nil, mustOpenFilePart(pw.ID(),
tst.root, tst.fileSystem, nil))
Review Comment:
When reopening flushed parts in `flush`, `mustOpenFilePart` is called with
`nil` cache. To populate and reuse cached metadata, it should pass `tst.cache`
instead of `nil`.
```suggestion
newPW := newPartWrapper(nil, mustOpenFilePart(pw.ID(),
tst.root, tst.fileSystem, tst.cache))
```
##########
banyand/measure/cache.go:
##########
@@ -0,0 +1,163 @@
+// Licensed to Apache Software Foundation (ASF) under one or more contributor
+// license agreements. See the NOTICE file distributed with
+// this work for additional information regarding copyright
+// ownership. Apache Software Foundation (ASF) licenses this file to you under
+// the Apache License, Version 2.0 (the "License"); you may
+// not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied. See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+package measure
+
+import (
+ "sync"
+ "sync/atomic"
+ "time"
+)
+
+const (
+ maxCachedBlocks = 100
+ cleanupPercentage = 0.1
+ cleanupInterval = 30 * time.Second
+ accessTimeout = 2 * time.Minute
+)
+
+type entry struct {
+ pbm []*primaryBlockMetadata
+ lastAccess uint64
+}
+
+type cache struct {
+ m map[uint64]*entry
+ cleanerStopCh chan struct{}
+ requests uint64
+ misses uint64
+ mu sync.RWMutex
+ cleanerWG sync.WaitGroup
+}
+
+func newCache() *cache {
+ c := &cache{
+ m: make(map[uint64]*entry),
+ cleanerStopCh: make(chan struct{}),
+ cleanerWG: sync.WaitGroup{},
+ }
+ c.cleanerWG.Add(1)
+ go func() {
+ defer c.cleanerWG.Done()
+ c.clean()
+ }()
+ return c
+}
+
+func (c *cache) clean() {
+ ticker := time.NewTicker(cleanupInterval)
+ defer ticker.Stop()
+ for {
+ select {
+ case <-ticker.C:
+ currentTime := uint64(time.Now().UnixNano())
+ c.mu.Lock()
+ for k, e := range c.m {
+ if currentTime-atomic.LoadUint64(&e.lastAccess)
> uint64(accessTimeout.Nanoseconds()) {
+ delete(c.m, k)
+ }
+ }
+ c.mu.Unlock()
+ case <-c.cleanerStopCh:
+ return
+ }
+ }
+}
+
+func (c *cache) close() {
+ close(c.cleanerStopCh)
+ c.cleanerWG.Wait()
+ c.m = nil
+}
+
+func (c *cache) get(partID uint64) []primaryBlockMetadata {
+ atomic.AddUint64(&c.requests, 1)
+
+ c.mu.RLock()
+ e := c.m[partID]
+ c.mu.RUnlock()
+
+ if e != nil {
+ if atomic.LoadUint64(&e.lastAccess) !=
uint64(time.Now().UnixNano()) {
+ atomic.StoreUint64(&e.lastAccess,
uint64(time.Now().UnixNano()))
+ }
+ pbm := make([]primaryBlockMetadata, 0)
+ for _, m := range e.pbm {
+ pbm = append(pbm, *m)
+ }
+ return pbm
+ }
+
+ atomic.AddUint64(&c.misses, 1)
+ return nil
+}
+
+func (c *cache) put(partID uint64, pbm []primaryBlockMetadata) {
+ c.mu.Lock()
+ defer c.mu.Unlock()
+
+ if overflow := len(c.m) - maxCachedBlocks; overflow > 0 {
+ overflow = int(float64(len(c.m)) * cleanupPercentage)
+ for partID := range c.m {
+ delete(c.m, partID)
+ overflow--
+ if overflow == 0 {
+ break
+ }
+ }
+ }
+
+ pbmPtrs := make([]*primaryBlockMetadata, 0)
+ for _, m := range pbm {
+ pbmPtrs = append(pbmPtrs, &m)
Review Comment:
Loop variable `m` is re-used across iterations, so `&m` yields pointers to
the same address. Use the index to take pointers (`&pbm[i]`) to each element
instead.
```suggestion
for i := range pbm {
pbmPtrs = append(pbmPtrs, &pbm[i])
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscr...@skywalking.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
