/// <inheritdoc />
public GarbageCollectResult GarbageCollect(Func <byte[], byte[], bool> canCollect, string columnFamilyName = null, CancellationToken cancellationToken = default, byte[] startValue = null)
{
var gcResult = new GarbageCollectResult();
// The implementation below ignores batching and removes keys one by one
gcResult.BatchSize = 1;
var columnFamilyInfo = GetColumnFamilyInfo(columnFamilyName);
var columnFamilyHandleToUse = columnFamilyInfo.Handle;
using (var iterator = m_store.NewIterator(columnFamilyHandleToUse))
{
if (startValue != null)
{
iterator.Seek(startValue);
}
else
{
iterator.SeekToFirst();
}
while (iterator.Valid() && !cancellationToken.IsCancellationRequested)
{
var startTime = TimestampUtilities.Timestamp;
gcResult.TotalCount++;
var bytesKey = iterator.Key();
var bytesValue = iterator.Value();
gcResult.LastKey = bytesKey;
if (canCollect(bytesKey, bytesValue))
{
Remove(bytesKey, columnFamilyName);
gcResult.RemovedCount++;
}
iterator.Next();
var duration = TimestampUtilities.Timestamp - startTime;
if (duration > gcResult.MaxBatchEvictionTime)
{
gcResult.MaxBatchEvictionTime = duration;
}
}
}
return(gcResult);
}
/// <inheritdoc />
public GarbageCollectResult GarbageCollectByKeyValue(
Func <Iterator, bool> canCollect,
string columnFamilyName = null,
IEnumerable <string> additionalColumnFamilies = null,
CancellationToken cancellationToken = default,
byte[] startValue = null)
{
var gcStats = new GarbageCollectResult
{
BatchSize = GarbageCollectionBatchSize
};
var columnFamilyInfo = GetColumnFamilyInfo(columnFamilyName);
var columnFamilyHandleToUse = columnFamilyInfo.Handle;
// According to RocksDB documentation, an iterator always loads both the key
// and the value. To avoid this, when possible, we use the key-tracked column with just keys
// and empty values and use that for eviction to avoid the load cost of full content.
if (columnFamilyInfo.UseKeyTracking)
{
columnFamilyHandleToUse = columnFamilyInfo.KeyHandle;
}
var keysToRemove = new List <byte[]>();
var primaryColumn = new string[] { columnFamilyName };
var columnsToUse = additionalColumnFamilies == null ? primaryColumn : additionalColumnFamilies.Concat(primaryColumn);
using (Iterator iterator = m_store.NewIterator(columnFamilyHandleToUse, m_readOptions))
{
if (startValue != null)
{
iterator.Seek(startValue);
}
else
{
iterator.SeekToFirst();
}
bool reachedEnd = !iterator.Valid();
while (!reachedEnd && !cancellationToken.IsCancellationRequested)
{
gcStats.TotalCount++;
bool canCollectResult = canCollect(iterator);
if (canCollectResult)
{
var bytesKey = iterator.Key();
keysToRemove.Add(bytesKey);
}
iterator.Next();
reachedEnd = !iterator.Valid();
if (keysToRemove.Count == GarbageCollectionBatchSize ||
(reachedEnd && keysToRemove.Count > 0))
{
var startTime = TimestampUtilities.Timestamp;
// Remove the key across all specified columns
RemoveBatch(keysToRemove, columnFamilyNames: columnsToUse);
var duration = TimestampUtilities.Timestamp - startTime;
if (duration > gcStats.MaxBatchEvictionTime)
{
gcStats.MaxBatchEvictionTime = duration;
}
gcStats.LastKey = keysToRemove.Last();
gcStats.RemovedCount += keysToRemove.Count;
keysToRemove.Clear();
}
}
}
gcStats.Canceled = cancellationToken.IsCancellationRequested;
return(gcStats);
}