Mostafa Mokhtar created IMPALA-6425:
---------------------------------------

             Summary: Change Mempool memory allocation size to be <1MB to avoid 
allocating from CentralFreeList
                 Key: IMPALA-6425
                 URL: https://issues.apache.org/jira/browse/IMPALA-6425
             Project: IMPALA
          Issue Type: Improvement
            Reporter: Mostafa Mokhtar
            Assignee: Tim Armstrong


While [~tlipcon] was investigating KRPC contention he noticed that 
MemPool::Allocate is doing 1MB allocations, which is somewhat of an 
anti-pattern with tcmalloc. 
During the tests MemPool was doing several thousand 1MB allocs per second and 
those have to do a full scan of the tcmalloc span linked list, which is very 
slow and only gets slower
 1040384 bytes on the other hand is constant time. 
 
It is not clear if a Power of 2 allocation size would help, worth experimenting 
with 512KB and 1040384 bytes. 
{code}

 /// The maximum size of chunk that should be allocated. Allocations larger 
than this
 /// size will get their own individual chunk.
 static const int MAX_CHUNK_SIZE = 8192*127

{code}

 

{code}

#0  0x0000000002097407 in base::internal::SpinLockDelay(int volatile*, int, 
int) ()
#1  0x00000000020e2049 in SpinLock::SlowLock() ()
#2  0x0000000002124348 in tcmalloc::CentralFreeList::Populate() ()
#3  0x0000000002124458 in tcmalloc::CentralFreeList::FetchFromOneSpansSafe(int, 
void**, void**) ()
#4  0x00000000021244e8 in tcmalloc::CentralFreeList::RemoveRange(void**, 
void**, int) ()
#5  0x0000000002131ee5 in tcmalloc::ThreadCache::FetchFromCentralCache(unsigned 
int, int, void* (*)(unsigned long)) ()
#6  0x0000000000b2879a in impala::MemPool::FindChunk(long, bool) ()
#7  0x0000000000b364f6 in impala::MemPool::Allocate(long) ()
#8  0x0000000000b36674 in impala::FreePool::Allocate(long) ()
#9  0x0000000000b353db in impala::RowBatch::Deserialize(kudu::Slice const&, 
kudu::Slice const&, long, bool, impala::FreePool*) ()
#10 0x0000000000b35795 in impala::RowBatch::RowBatch(impala::RowDescriptor 
const*, impala::RowBatchHeaderPB const&, kudu::Slice const&, kudu::Slice 
const&, impala::FreePool*) ()
#11 0x0000000000b1644f in 
impala::KrpcDataStreamRecvr::SenderQueue::AddBatchWork(long, 
impala::RowBatchHeaderPB const&, kudu::Slice const&, kudu::Slice const&, 
boost::unique_lock<impala::SpinLock>*) ()
#12 0x0000000000b19135 in 
impala::KrpcDataStreamRecvr::SenderQueue::AddBatch(impala::TransmitDataRequestPB
 const*, kudu::rpc::RpcContext*) ()
#13 0x0000000000b0ee30 in 
impala::KrpcDataStreamMgr::AddData(impala::TransmitDataRequestPB const*, 
kudu::rpc::RpcContext*) ()
#14 0x0000000001187035 in 
kudu::rpc::GeneratedServiceIf::Handle(kudu::rpc::InboundCall*) ()
#15 0x00000000011bc1cd in impala::ImpalaServicePool::RunThread(long) ()

{code}

 

Also it appears that the thread above was a victim of thread below, yet 
allocations <1MB will make MemPool::Allocate content less over the 
CentralFreeList lock. 

{code}

#0 0x0000003173ae5407 in madvise () from /lib64/libc.so.6
#1 0x0000000002131cca in TCMalloc_SystemRelease(void*, unsigned long) ()
#2 0x000000000212f26a in tcmalloc::PageHeap::DecommitSpan(tcmalloc::Span*) ()
#3 0x000000000212f505 in tcmalloc::PageHeap::MergeIntoFreeList(tcmalloc::Span*) 
()
#4 0x000000000212f864 in tcmalloc::PageHeap::Delete(tcmalloc::Span*) ()
#5 0x0000000002123cf7 in tcmalloc::CentralFreeList::ReleaseToSpans(void*) ()
#6 0x0000000002123d9b in tcmalloc::CentralFreeList::ReleaseListToSpans(void*) ()
#7 0x0000000002124067 in tcmalloc::CentralFreeList::InsertRange(void*, void*, 
int) ()
#8 0x00000000021320a4 in 
tcmalloc::ThreadCache::ReleaseToCentralCache(tcmalloc::ThreadCache::FreeList*, 
unsigned int, int) ()
#9 0x0000000002132575 in 
tcmalloc::ThreadCache::ListTooLong(tcmalloc::ThreadCache::FreeList*, unsigned 
int) ()
#10 0x0000000000b276e0 in impala::MemPool::FreeAll() ()
#11 0x0000000000b34655 in impala::RowBatch::Reset() ()
#12 0x0000000000fe882f in 
impala::PartitionedAggregationNode::GetRowsStreaming(impala::RuntimeState*, 
impala::RowBatch*) ()
#13 0x0000000000fe9771 in 
impala::PartitionedAggregationNode::GetNext(impala::RuntimeState*, 
impala::RowBatch*, bool*) ()
#14 0x0000000000b78352 in impala::FragmentInstanceState::ExecInternal() ()
#15 0x0000000000b7adc2 in impala::FragmentInstanceState::Exec() ()
#16 0x0000000000b6a0da in 
impala::QueryState::ExecFInstance(impala::FragmentInstanceState*) ()

{code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to