/// <summary>Scan through the cached block map.</summary>
/// <remarks>
/// Scan through the cached block map.
/// Any blocks which are under-replicated should be assigned new Datanodes.
/// Blocks that are over-replicated should be removed from Datanodes.
/// </remarks>
private void RescanCachedBlockMap()
{
for (IEnumerator <CachedBlock> cbIter = cachedBlocks.GetEnumerator(); cbIter.HasNext
();)
{
scannedBlocks++;
CachedBlock cblock = cbIter.Next();
IList <DatanodeDescriptor> pendingCached = cblock.GetDatanodes(DatanodeDescriptor.CachedBlocksList.Type
.PendingCached);
IList <DatanodeDescriptor> cached = cblock.GetDatanodes(DatanodeDescriptor.CachedBlocksList.Type
.Cached);
IList <DatanodeDescriptor> pendingUncached = cblock.GetDatanodes(DatanodeDescriptor.CachedBlocksList.Type
.PendingUncached);
// Remove nodes from PENDING_UNCACHED if they were actually uncached.
for (IEnumerator <DatanodeDescriptor> iter = pendingUncached.GetEnumerator(); iter
.HasNext();)
{
DatanodeDescriptor datanode = iter.Next();
if (!cblock.IsInList(datanode.GetCached()))
{
Log.Trace("Block {}: removing from PENDING_UNCACHED for node {} " + "because the DataNode uncached it."
, cblock.GetBlockId(), datanode.GetDatanodeUuid());
datanode.GetPendingUncached().Remove(cblock);
iter.Remove();
}
}
BlockInfoContiguous blockInfo = blockManager.GetStoredBlock(new Block(cblock.GetBlockId
()));
string reason = FindReasonForNotCaching(cblock, blockInfo);
int neededCached = 0;
if (reason != null)
{
Log.Trace("Block {}: can't cache block because it is {}", cblock.GetBlockId(), reason
);
}
else
{
neededCached = cblock.GetReplication();
}
int numCached = cached.Count;
if (numCached >= neededCached)
{
// If we have enough replicas, drop all pending cached.
for (IEnumerator <DatanodeDescriptor> iter_1 = pendingCached.GetEnumerator(); iter_1
.HasNext();)
{
DatanodeDescriptor datanode = iter_1.Next();
datanode.GetPendingCached().Remove(cblock);
iter_1.Remove();
Log.Trace("Block {}: removing from PENDING_CACHED for node {}" + "because we already have {} cached replicas and we only"
+ " need {}", cblock.GetBlockId(), datanode.GetDatanodeUuid(), numCached, neededCached
);
}
}
if (numCached < neededCached)
{
// If we don't have enough replicas, drop all pending uncached.
for (IEnumerator <DatanodeDescriptor> iter_1 = pendingUncached.GetEnumerator(); iter_1
.HasNext();)
{
DatanodeDescriptor datanode = iter_1.Next();
datanode.GetPendingUncached().Remove(cblock);
iter_1.Remove();
Log.Trace("Block {}: removing from PENDING_UNCACHED for node {} " + "because we only have {} cached replicas and we need "
+ "{}", cblock.GetBlockId(), datanode.GetDatanodeUuid(), numCached, neededCached
);
}
}
int neededUncached = numCached - (pendingUncached.Count + neededCached);
if (neededUncached > 0)
{
AddNewPendingUncached(neededUncached, cblock, cached, pendingUncached);
}
else
{
int additionalCachedNeeded = neededCached - (numCached + pendingCached.Count);
if (additionalCachedNeeded > 0)
{
AddNewPendingCached(additionalCachedNeeded, cblock, cached, pendingCached);
}
}
if ((neededCached == 0) && pendingUncached.IsEmpty() && pendingCached.IsEmpty())
{
// we have nothing more to do with this block.
Log.Trace("Block {}: removing from cachedBlocks, since neededCached " + "== 0, and pendingUncached and pendingCached are empty."
, cblock.GetBlockId());
cbIter.Remove();
}
}
}
/// <summary>Apply a CacheDirective to a file.</summary>
/// <param name="directive">The CacheDirective to apply.</param>
/// <param name="file">The file.</param>
private void RescanFile(CacheDirective directive, INodeFile file)
{
BlockInfoContiguous[] blockInfos = file.GetBlocks();
// Increment the "needed" statistics
directive.AddFilesNeeded(1);
// We don't cache UC blocks, don't add them to the total here
long neededTotal = file.ComputeFileSizeNotIncludingLastUcBlock() * directive.GetReplication
();
directive.AddBytesNeeded(neededTotal);
// The pool's bytesNeeded is incremented as we scan. If the demand
// thus far plus the demand of this file would exceed the pool's limit,
// do not cache this file.
CachePool pool = directive.GetPool();
if (pool.GetBytesNeeded() > pool.GetLimit())
{
Log.Debug("Directive {}: not scanning file {} because " + "bytesNeeded for pool {} is {}, but the pool's limit is {}"
, directive.GetId(), file.GetFullPathName(), pool.GetPoolName(), pool.GetBytesNeeded
(), pool.GetLimit());
return;
}
long cachedTotal = 0;
foreach (BlockInfoContiguous blockInfo in blockInfos)
{
if (!blockInfo.GetBlockUCState().Equals(HdfsServerConstants.BlockUCState.Complete
))
{
// We don't try to cache blocks that are under construction.
Log.Trace("Directive {}: can't cache block {} because it is in state " + "{}, not COMPLETE."
, directive.GetId(), blockInfo, blockInfo.GetBlockUCState());
continue;
}
Block block = new Block(blockInfo.GetBlockId());
CachedBlock ncblock = new CachedBlock(block.GetBlockId(), directive.GetReplication
(), mark);
CachedBlock ocblock = cachedBlocks.Get(ncblock);
if (ocblock == null)
{
cachedBlocks.Put(ncblock);
ocblock = ncblock;
}
else
{
// Update bytesUsed using the current replication levels.
// Assumptions: we assume that all the blocks are the same length
// on each datanode. We can assume this because we're only caching
// blocks in state COMPLETE.
// Note that if two directives are caching the same block(s), they will
// both get them added to their bytesCached.
IList <DatanodeDescriptor> cachedOn = ocblock.GetDatanodes(DatanodeDescriptor.CachedBlocksList.Type
.Cached);
long cachedByBlock = Math.Min(cachedOn.Count, directive.GetReplication()) * blockInfo
.GetNumBytes();
cachedTotal += cachedByBlock;
if ((mark != ocblock.GetMark()) || (ocblock.GetReplication() < directive.GetReplication
()))
{
//
// Overwrite the block's replication and mark in two cases:
//
// 1. If the mark on the CachedBlock is different from the mark for
// this scan, that means the block hasn't been updated during this
// scan, and we should overwrite whatever is there, since it is no
// longer valid.
//
// 2. If the replication in the CachedBlock is less than what the
// directive asks for, we want to increase the block's replication
// field to what the directive asks for.
//
ocblock.SetReplicationAndMark(directive.GetReplication(), mark);
}
}
Log.Trace("Directive {}: setting replication for block {} to {}", directive.GetId
(), blockInfo, ocblock.GetReplication());
}
// Increment the "cached" statistics
directive.AddBytesCached(cachedTotal);
if (cachedTotal == neededTotal)
{
directive.AddFilesCached(1);
}
Log.Debug("Directive {}: caching {}: {}/{} bytes", directive.GetId(), file.GetFullPathName
(), cachedTotal, neededTotal);
}