private void MaybeAddFutureBatch(List<JsonDocument> past)
{
if (context.Configuration.DisableDocumentPreFetchingForIndexing || context.RunIndexing == false)
return;
if (context.Configuration.MaxNumberOfParallelIndexTasks == 1)
return;
if (past.Count == 0)
return;
if (futureIndexBatches.Count > 5) // we limit the number of future calls we do
{
int alreadyLoaded = futureIndexBatches.Values.Sum(x =>
{
if (x.Task.IsCompleted)
return x.Task.Result.Count;
return autoTuner.NumberOfItemsToIndexInSingleBatch / 4 * 3;
});
if (alreadyLoaded > autoTuner.NumberOfItemsToIndexInSingleBatch)
return;
}
// ensure we don't do TOO much future caching
if (MemoryStatistics.AvailableMemory <
context.Configuration.AvailableMemoryForRaisingIndexBatchSizeLimit)
return;
// we loaded the maximum amount, there are probably more items to read now.
Etag highestLoadedEtag = GetHighestEtag(past);
Etag nextEtag = GetNextDocumentEtagFromDisk(highestLoadedEtag);
if (nextEtag == highestLoadedEtag)
return; // there is nothing newer to do
if (futureIndexBatches.ContainsKey(nextEtag)) // already loading this
return;
var futureBatchStat = new FutureBatchStats
{
Timestamp = SystemTime.UtcNow,
};
Stopwatch sp = Stopwatch.StartNew();
context.AddFutureBatch(futureBatchStat);
futureIndexBatches.TryAdd(nextEtag, new FutureIndexBatch
{
StartingEtag = nextEtag,
Age = Interlocked.Increment(ref currentIndexingAge),
Task = Task.Factory.StartNew(() =>
{
List<JsonDocument> jsonDocuments = null;
int localWork = 0;
while (context.RunIndexing)
{
jsonDocuments = GetJsonDocsFromDisk(EtagUtil.Increment(nextEtag, -1), null);
if (jsonDocuments.Count > 0)
break;
futureBatchStat.Retries++;
context.WaitForWork(TimeSpan.FromMinutes(10), ref localWork, "PreFetching");
}
futureBatchStat.Duration = sp.Elapsed;
futureBatchStat.Size = jsonDocuments == null ? 0 : jsonDocuments.Count;
if (jsonDocuments != null)
{
MaybeAddFutureBatch(jsonDocuments);
}
return jsonDocuments;
})
});
}
private void MaybeAddFutureBatch(JsonResults past)
{
if (context.Configuration.MaxNumberOfParallelIndexTasks == 1)
return;
if (past.Results.Length == 0 || past.LoadedFromDisk == false)
return;
if (futureIndexBatches.Count > 5) // we limit the number of future calls we do
{
var alreadyLoaded = futureIndexBatches.Sum(x =>
{
if (x.Task.IsCompleted)
return x.Task.Result.Results.Length;
return 0;
});
if (alreadyLoaded > autoTuner.NumberOfItemsToIndexInSingleBatch)
return;
}
// ensure we don't do TOO much future cachings
if (MemoryStatistics.AvailableMemory < context.Configuration.AvailableMemoryForRaisingIndexBatchSizeLimit)
return;
// we loaded the maximum amount, there are probably more items to read now.
var nextEtag = GetNextHighestEtag(past.Results);
var nextBatch = futureIndexBatches.FirstOrDefault(x => x.StartingEtag == nextEtag);
if (nextBatch != null)
return;
var futureBatchStat = new FutureBatchStats
{
Timestamp = SystemTime.UtcNow,
};
var sp = Stopwatch.StartNew();
context.AddFutureBatch(futureBatchStat);
futureIndexBatches.Add(new FutureIndexBatch
{
StartingEtag = nextEtag,
Age = currentIndexingAge,
Task = System.Threading.Tasks.Task.Factory.StartNew(() =>
{
var jsonDocuments = GetJsonDocuments(nextEtag);
int localWork = workCounter;
while (jsonDocuments.Results.Length == 0 && context.DoWork)
{
futureBatchStat.Retries++;
if (context.WaitForWork(TimeSpan.FromMinutes(10), ref localWork, "PreFetching") == false)
continue;
jsonDocuments = GetJsonDocuments(nextEtag);
}
futureBatchStat.Duration = sp.Elapsed;
futureBatchStat.Size = jsonDocuments.Results.Length;
MaybeAddFutureBatch(jsonDocuments);
return jsonDocuments;
})
});
}
public void AddFutureBatch(FutureBatchStats futureBatchStat)
{
futureBatchStats.Add(futureBatchStat);
if (futureBatchStats.Count <= 30)
return;
foreach (var source in futureBatchStats.OrderBy(x => x.Timestamp).Take(5))
{
futureBatchStats.TryRemove(source);
}
}
private bool AddFutureBatch(Etag nextEtag, Etag untilEtag, bool isSplitted = false, bool isEarlyExitBatch = false)
{
var futureBatchStat = new FutureBatchStats
{
Timestamp = SystemTime.UtcNow,
PrefetchingUser = PrefetchingUser
};
Stopwatch sp = Stopwatch.StartNew();
context.AddFutureBatch(futureBatchStat);
var cts = new CancellationTokenSource();
var linkedToken = CancellationTokenSource.CreateLinkedTokenSource(cts.Token, context.CancellationToken);
var futureIndexBatch = new FutureIndexBatch
{
StartingEtag = nextEtag,
Age = Interlocked.Increment(ref currentIndexingAge),
CancellationTokenSource = linkedToken,
IsSplitted = isSplitted,
Task = Task.Factory.StartNew(() =>
{
List<JsonDocument> jsonDocuments = null;
int localWork = 0;
var earlyExit = new Reference<bool>();
while (context.RunIndexing)
{
linkedToken.Token.ThrowIfCancellationRequested();
jsonDocuments = GetJsonDocsFromDisk(
linkedToken.Token,
Abstractions.Util.EtagUtil.Increment(nextEtag, -1), untilEtag, earlyExit);
if (jsonDocuments.Count > 0)
break;
futureBatchStat.Retries++;
context.WaitForWork(TimeSpan.FromMinutes(10), ref localWork, "PreFetching");
}
futureBatchStat.Duration = sp.Elapsed;
futureBatchStat.Size = jsonDocuments == null ? 0 : jsonDocuments.Count;
if (jsonDocuments == null)
return null;
LogEarlyExit(nextEtag, untilEtag, isEarlyExitBatch, jsonDocuments, sp.ElapsedMilliseconds);
if (untilEtag != null && earlyExit.Value)
{
var lastEtag = GetHighestEtag(jsonDocuments);
context.TransactionalStorage.Batch(accessor =>
{
lastEtag = accessor.Documents.GetBestNextDocumentEtag(lastEtag);
});
if (log.IsDebugEnabled)
{
log.Debug("Early exit from last future splitted batch, need to fetch documents from etag: {0} to etag: {1}",
lastEtag, untilEtag);
}
linkedToken.Token.ThrowIfCancellationRequested();
AddFutureBatch(lastEtag, untilEtag, isEarlyExitBatch: true);
}
else
{
linkedToken.Token.ThrowIfCancellationRequested();
MaybeAddFutureBatch(jsonDocuments);
}
return jsonDocuments;
}, linkedToken.Token)
.ContinueWith(t =>
{
t.AssertNotFailed();
linkedToken = null;
return t.Result;
}, linkedToken.Token)
};
futureIndexBatch.Task.ContinueWith(t =>
{
FutureBatchCompleted(t.Result.Count);
}, linkedToken.Token);
return futureIndexBatches.TryAdd(nextEtag, futureIndexBatch);
}
private bool AddFutureBatch(Etag nextEtag, Etag untilEtag,
FutureBatchType batchType, int? docsCount = null)
{
var futureBatchStat = new FutureBatchStats
{
Timestamp = SystemTime.UtcNow,
PrefetchingUser = PrefetchingUser
};
var sp = Stopwatch.StartNew();
context.AddFutureBatch(futureBatchStat);
var docsCountRef = new Reference<int?>() {Value = docsCount};
var cts = new CancellationTokenSource();
var linkedToken = CancellationTokenSource.CreateLinkedTokenSource(cts.Token, context.CancellationToken);
var futureIndexBatch = new FutureIndexBatch
{
StartingEtag = nextEtag,
Age = Interlocked.Increment(ref currentIndexingAge),
CancellationTokenSource = cts,
Type = batchType,
DocsCount = docsCountRef,
Task = Task.Run(() =>
{
List<JsonDocument> jsonDocuments = null;
int localWork = 0;
var earlyExit = new Reference<bool>();
while (context.RunIndexing)
{
linkedToken.Token.ThrowIfCancellationRequested();
jsonDocuments = GetJsonDocsFromDisk(
linkedToken.Token,
Abstractions.Util.EtagUtil.Increment(nextEtag, -1), untilEtag, earlyExit);
if (jsonDocuments.Count > 0)
break;
futureBatchStat.Retries++;
context.WaitForWork(TimeSpan.FromMinutes(10), ref localWork, "PreFetching");
}
futureBatchStat.Duration = sp.Elapsed;
futureBatchStat.Size = jsonDocuments == null ? 0 : jsonDocuments.Count;
if (jsonDocuments == null)
return null;
LogEarlyExit(nextEtag, untilEtag, batchType == FutureBatchType.EarlyExit,
jsonDocuments, sp.ElapsedMilliseconds);
if (untilEtag != null && earlyExit.Value)
{
var lastEtag = GetHighestEtag(jsonDocuments);
context.TransactionalStorage.Batch(accessor =>
{
lastEtag = accessor.Documents.GetBestNextDocumentEtag(lastEtag);
});
if (log.IsDebugEnabled)
{
log.Debug("Early exit from last future splitted batch, need to fetch documents from etag: {0} to etag: {1}",
lastEtag, untilEtag);
}
linkedToken.Token.ThrowIfCancellationRequested();
docsCountRef.Value = jsonDocuments.Count;
var docsLeft = docsCount - jsonDocuments.Count;
if (docsLeft > 0 && lastEtag.CompareTo(untilEtag) <= 0)
AddFutureBatch(lastEtag, untilEtag, FutureBatchType.EarlyExit, docsLeft);
}
else
{
linkedToken.Token.ThrowIfCancellationRequested();
MaybeAddFutureBatch(jsonDocuments);
}
return jsonDocuments;
}, linkedToken.Token)
.ContinueWith(t =>
{
using (cts)
using (linkedToken)
{
t.AssertNotFailed();
}
return t.Result;
})
};
futureIndexBatch.Task.ContinueWith(t =>
{
try
{
if (linkedToken.IsCancellationRequested == false)
FutureBatchCompleted(t.Result.Count);
}
catch (ObjectDisposedException)
{
// this is an expected race with the actual task, this is fine
}
});
var addFutureBatch = futureIndexBatches.TryAdd(nextEtag, futureIndexBatch);
if (addFutureBatch == false)
{
log.Info(string.Format("A future batch starting with {0} etag is already running", nextEtag));
cts.Cancel();
}
return addFutureBatch;
}
private void MaybeAddFutureBatch(List<JsonDocument> past)
{
if (context.Configuration.DisableDocumentPreFetching || context.RunIndexing == false)
return;
if (context.Configuration.MaxNumberOfParallelProcessingTasks == 1)
return;
if (past.Count == 0)
return;
if (prefetchingQueue.LoadedSize > autoTuner.MaximumSizeAllowedToFetchFromStorageInBytes)
return; // already have too much in memory
// don't keep _too_ much in memory
if (prefetchingQueue.Count > context.Configuration.MaxNumberOfItemsToProcessInSingleBatch * 2)
return;
var size = 1024;
var count = context.LastActualIndexingBatchInfo.Count;
if (count > 0)
{
size = context.LastActualIndexingBatchInfo.Aggregate(0, (o, c) => o + c.TotalDocumentCount)/count;
}
var alreadyLoadedSize = futureIndexBatches.Values.Sum(x =>
{
if (x.Task.IsCompleted)
return x.Task.Result.Sum(doc => doc.SerializedSizeOnDisk);
return size;
});
if (alreadyLoadedSize > context.Configuration.AvailableMemoryForRaisingBatchSizeLimit)
return;
if(MemoryStatistics.IsLowMemory)
return;
if (futureIndexBatches.Count > 5) // we limit the number of future calls we do
{
int alreadyLoaded = futureIndexBatches.Values.Sum(x =>
{
if (x.Task.IsCompleted)
return x.Task.Result.Count;
return autoTuner.NumberOfItemsToProcessInSingleBatch / 4 * 3;
});
if (alreadyLoaded > autoTuner.NumberOfItemsToProcessInSingleBatch)
return;
}
// ensure we don't do TOO much future caching
if (MemoryStatistics.AvailableMemory <
context.Configuration.AvailableMemoryForRaisingBatchSizeLimit)
return;
// we loaded the maximum amount, there are probably more items to read now.
Etag highestLoadedEtag = GetHighestEtag(past);
Etag nextEtag = GetNextDocumentEtagFromDisk(highestLoadedEtag);
if (nextEtag == highestLoadedEtag)
return; // there is nothing newer to do
if (futureIndexBatches.ContainsKey(nextEtag)) // already loading this
return;
var futureBatchStat = new FutureBatchStats
{
Timestamp = SystemTime.UtcNow,
PrefetchingUser = PrefetchingUser
};
Stopwatch sp = Stopwatch.StartNew();
context.AddFutureBatch(futureBatchStat);
futureIndexBatches.TryAdd(nextEtag, new FutureIndexBatch
{
StartingEtag = nextEtag,
Age = Interlocked.Increment(ref currentIndexingAge),
Task = Task.Factory.StartNew(() =>
{
List<JsonDocument> jsonDocuments = null;
int localWork = 0;
while (context.RunIndexing)
{
jsonDocuments = GetJsonDocsFromDisk(Abstractions.Util.EtagUtil.Increment(nextEtag, -1), null);
if (jsonDocuments.Count > 0)
break;
futureBatchStat.Retries++;
context.WaitForWork(TimeSpan.FromMinutes(10), ref localWork, "PreFetching");
}
futureBatchStat.Duration = sp.Elapsed;
futureBatchStat.Size = jsonDocuments == null ? 0 : jsonDocuments.Count;
if (jsonDocuments != null)
{
MaybeAddFutureBatch(jsonDocuments);
}
return jsonDocuments;
})
});
}
