You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@impala.apache.org by meijie <me...@yandex.com> on 2018/08/02 07:31:37 UTC

Slow sql because NumScannerThreadsStarted is 1

Hi, Every One

I found the sql (select * from parquet_session_1 where browser = 
'Firefox 20') was very slow at my impala cluster. which needed 15 
seconds to executed and NumScannerThreadsStarted is 1 , it wasn't as 
expected.

the part profile as blew:

            - FooterProcessingTime: (Avg: 4.430ms ; Min: 2.924ms ; Max: 
9.559ms ; Number of samples: 9)
            - AverageHdfsReadThreadConcurrency: 0.00
            - AverageScannerThreadConcurrency: 1.00
            - BytesRead: 2.13 GB (2288275802)
            - BytesReadDataNodeCache: 0
            - BytesReadLocal: 2.13 GB (2288275802)
            - BytesReadRemoteUnexpected: 0
            - BytesReadShortCircuit: 2.13 GB (2288275802)
            - CachedFileHandlesHitCount: 0 (0)
            - CachedFileHandlesMissCount: 171 (171)
            - CollectionItemsRead: 0 (0)
            - DecompressionTime: 3s264ms
            - MaxCompressedTextFileLength: 0
            - NumColumns: 18 (18)
            - NumDictFilteredRowGroups: 0 (0)
            - NumDisksAccessed: 5 (5)
            - NumRowGroups: 9 (9)
            - NumScannerThreadsStarted: 1 (1)
            - NumScannersWithNoReads: 0 (0)
            - NumStatsFilteredRowGroups: 0 (0)
            - PeakMemoryUsage: 143.25 MB (150207517)
            - PerReadThreadRawHdfsThroughput: 1.30 GB/sec
            - RemoteScanRanges: 0 (0)
            - RowBatchQueueGetWaitTime: 14s216ms
            - RowBatchQueuePutWaitTime: 0.000ns
            - RowsRead: 51.90M (51903646)
            - RowsReturned: 93 (93)
            - RowsReturnedRate: 6.00 /sec
            - ScanRangesComplete: 9 (9)
            - ScannerThreadsInvoluntaryContextSwitches: 18 (18)
            - ScannerThreadsTotalWallClockTime: 14s678ms
              - MaterializeTupleTime(*): 10s639ms
              - ScannerThreadsSysTime: 698.325ms
              - ScannerThreadsUserTime: 13s452ms
            - ScannerThreadsVoluntaryContextSwitches: 1.38K (1382)
            - TotalRawHdfsReadTime(*): 1s637ms
            - TotalReadThroughput: 148.73 MB/sec

But, if I only select one column(select uuid from parquet_session_1 
where browser = 'Firefox 20'), It just needed 0.5 seconds and 
NumScannerThreadsStarted is 4. I don't know what affects the number of 
scan threads.

Impala Version: 2.11.0-cdh5.14.0 RELEASE, the sql submitted by impala-shell

Thanks

-- 
Best Regard
meijie


Re: Slow sql because NumScannerThreadsStarted is 1

Posted by Tim Armstrong <ta...@cloudera.com>.
Thanks, that helps!

There are a number of reasons why more scanner threads may not be started:

   - Not enough parallelism possible (e.g. only scanning one Parquet file)
   - Other concurrent queries using many threads (we have a soft limit of 3
   * number of logical cores threads per impalad)
   - Not enough memory to start a new thread (based on an estimate of
   memory consumption

I think in your case it's probably memory - the estimate for Parquet is 3 *
8MB * <number of columns scanned>, so 432MB. Your query's memory limit is
400MB so it will think that it can't start a second thread without running
out of memory.

That's probably a bit too conservative for your use case. You could speed
it up by reducing the number of columns scanned (i.e. replace select * with
a shorter list) or by increasing the mem_limit for the query.

By coincidence I'm working on refining this heuristic for our next 3.1
release, since it's often too conservative.

On Thu, Aug 2, 2018 at 7:34 PM, meijie <me...@yandex.com> wrote:

> Thank you for your response. blew is complete profile
>
> Query Runtime Profile:
> Query (id=b4f4f6493e9fe5a:fc9961d600000000):
>   Summary:
>     Session ID: ef47d3119c267424:d9df967cefd5a883
>     Session Type: BEESWAX
>     Start Time: 2018-08-03 09:38:45.127171000
>     End Time: 2018-08-03 09:39:02.285440000
>     Query Type: QUERY
>     Query State: FINISHED
>     Query Status: OK
>     Impala Version: impalad version 2.11.0-cdh5.14.0 RELEASE (build
> d68206561bce6b26762d62c01a78e6cd27aa7690)
>     User: root
>     Connected User: root
>     Delegated User:
>     Network Address: ::ffff:172.16.20.50:55138
>     Default Db: bigdb_press_test
>     Sql Statement: select * from parquet_session_1 where browser =
> 'Firefox 20'
>     Coordinator: bigdb01:22000
>     Query Options (set by configuration): MEM_LIMIT=419430400
>     Query Options (set by configuration and planner):
> MEM_LIMIT=419430400,MT_DOP=0
>     Plan:
> ----------------
> Max Per-Host Resource Reservation: Memory=0B
> Per-Host Resource Estimates: Memory=1.55GB
>
> F01:PLAN FRAGMENT [UNPARTITIONED] hosts=1 instances=1
> |  Per-Host Resources: mem-estimate=0B mem-reservation=0B
> PLAN-ROOT SINK
> |  mem-estimate=0B mem-reservation=0B
> |
> 01:EXCHANGE [UNPARTITIONED]
> |  mem-estimate=0B mem-reservation=0B
> |  tuple-ids=0 row-size=396B cardinality=1140504
> |
> F00:PLAN FRAGMENT [RANDOM] hosts=3 instances=3
> Per-Host Resources: mem-estimate=1.55GB mem-reservation=0B
> 00:SCAN HDFS [bigdb_press_test.parquet_session_1, RANDOM]
>    partitions=1/1 files=27 size=6.37GB
>    predicates: browser = 'Firefox 20'
>    stats-rows=155108551 extrapolated-rows=disabled
>    table stats: rows=155108551 size=6.37GB
>    column stats: all
>    parquet statistics predicates: browser = 'Firefox 20'
>    parquet dictionary predicates: browser = 'Firefox 20'
>    mem-estimate=1.55GB mem-reservation=0B
>    tuple-ids=0 row-size=396B cardinality=1140504
> ----------------
>     Estimated Per-Host Mem: 1660944384
>     Per Host Min Reservation: bigdb01:22000(0) bigdb02:22000(0)
> bigdb03:22000(0)
>     Request Pool: root.default
>     Admission result: Admitted immediately
>     ExecSummary:
> Operator       #Hosts  Avg Time  Max Time  #Rows  Est. #Rows   Peak Mem
> Est. Peak Mem  Detail
> ------------------------------------------------------------
> ---------------------------------------------------------------
> 01:EXCHANGE         1  16s927ms  16s927ms    286       1.14M
> 0              0  UNPARTITIONED
> 00:SCAN HDFS        3  14s876ms  16s448ms    286       1.14M  143.25
> MB        1.55 GB  bigdb_press_test.parquet_session_1
>     Errors:
>     Planner Timeline: 4.874ms
>        - Analysis finished: 1.479ms (1.479ms)
>        - Value transfer graph computed: 1.694ms (215.699us)
>        - Single node plan created: 3.140ms (1.445ms)
>        - Runtime filters computed: 3.256ms (116.502us)
>        - Distributed plan created: 3.324ms (67.434us)
>        - Lineage info computed: 3.467ms (143.587us)
>        - Planning finished: 4.874ms (1.406ms)
>     Query Timeline: 17s161ms
>        - Query submitted: 75.337us (75.337us)
>        - Planning finished: 6.136ms (6.061ms)
>        - Submit for admission: 6.636ms (500.206us)
>        - Completed admission: 6.716ms (79.190us)
>        - Ready to start on 3 backends: 6.935ms (219.689us)
>        - All 3 execution backends (4 fragment instances) started: 8.842ms
> (1.906ms)
>        - Rows available: 351.212ms (342.370ms)
>        - First row fetched: 441.198ms (89.985ms)
>        - Unregister query: 17s158ms (16s717ms)
>      - ComputeScanRangeAssignmentTimer: 109.935us
>   ImpalaServer:
>      - ClientFetchWaitTimer: 280.979ms
>      - RowMaterializationTimer: 16s526ms
>   Execution Profile b4f4f6493e9fe5a:fc9961d600000000:(Total: 16s870ms,
> non-child: 0.000ns, % non-child: 0.00%)
>     Number of filters: 0
>     Filter routing table:
>  ID  Src. Node  Tgt. Node(s)  Target type  Partition filter  Pending
> (Expected)  First arrived  Completed   Enabled
> ------------------------------------------------------------
> -------------------------------------------------------
>
>     Backend startup latencies: Count: 3, min / max: 1ms / 2ms, 25th %-ile:
> 1ms, 50th %-ile: 2ms, 75th %-ile: 2ms, 90th %-ile: 2ms, 95th %-ile: 2ms,
> 99.9th %-ile: 2ms
>     Per Node Peak Memory Usage: bigdb01:22000(144.57 MB)
> bigdb02:22000(144.09 MB) bigdb03:22000(144.09 MB)
>      - FiltersReceived: 0 (0)
>      - FinalizationTimer: 0.000ns
>      - NumBackends: 3 (3)
>      - NumFragmentInstances: 4 (4)
>      - NumFragments: 2 (2)
>     Averaged Fragment F01:(Total: 16s955ms, non-child: 2.075ms, %
> non-child: 0.01%)
>       split sizes:  min: 0, max: 0, avg: 0, stddev: 0
>       completion times: min:16s957ms  max:16s957ms  mean: 16s957ms
> stddev:0.000ns
>       execution rates: min:0.00 /sec  max:0.00 /sec  mean:0.00 /sec
> stddev:0.00 /sec
>       num instances: 1
>        - AverageThreadTokens: 0.00
>        - BloomFilterBytes: 0
>        - PeakMemoryUsage: 580.02 KB (593936)
>        - PeakReservation: 0
>        - PeakUsedReservation: 0
>        - PerHostPeakMemUsage: 144.57 MB (151595498)
>        - RowsProduced: 286 (286)
>        - TotalNetworkReceiveTime: 16s926ms
>        - TotalNetworkSendTime: 0.000ns
>        - TotalStorageWaitTime: 0.000ns
>        - TotalThreadsInvoluntaryContextSwitches: 0 (0)
>        - TotalThreadsTotalWallClockTime: 16s929ms
>          - TotalThreadsSysTime: 849.000us
>          - TotalThreadsUserTime: 3.925ms
>        - TotalThreadsVoluntaryContextSwitches: 144 (144)
>       Fragment Instance Lifecycle Timings:
>          - ExecTime: 16s614ms
>            - ExecTreeExecTime: 16s612ms
>          - OpenTime: 314.632ms
>            - ExecTreeOpenTime: 314.614ms
>          - PrepareTime: 26.668ms
>            - ExecTreePrepareTime: 43.413us
>       PLAN_ROOT_SINK:
>          - PeakMemoryUsage: 0
>       CodeGen:(Total: 26.326ms, non-child: 26.326ms, % non-child: 100.00%)
>          - CodegenTime: 0.000ns
>          - CompileTime: 0.000ns
>          - LoadTime: 0.000ns
>          - ModuleBitcodeSize: 1.86 MB (1950884)
>          - NumFunctions: 0 (0)
>          - NumInstructions: 0 (0)
>          - OptimizationTime: 0.000ns
>          - PeakMemoryUsage: 0
>          - PrepareTime: 25.464ms
>       EXCHANGE_NODE (id=1):(Total: 16s927ms, non-child: 16s927ms, %
> non-child: 100.00%)
>          - ConvertRowBatchTime: 290.044us
>          - PeakMemoryUsage: 0
>          - RowsReturned: 286 (286)
>          - RowsReturnedRate: 16.00 /sec
>         DataStreamReceiver:
>            - BytesReceived: 54.30 KB (55602)
>            - DeserializeRowBatchTimer: 2.272ms
>            - FirstBatchArrivalWaitTime: 314.602ms
>            - PeakMemoryUsage: 4.09 KB (4184)
>            - SendersBlockedTimer: 0.000ns
>            - SendersBlockedTotalTimer(*): 0.000ns
>     Coordinator Fragment F01:
>       Instance b4f4f6493e9fe5a:fc9961d600000000
> (host=bigdb01:22000):(Total: 16s955ms, non-child: 2.075ms, % non-child:
> 0.01%)
>         MemoryUsage(500.000ms): 8.00 KB, 48.01 KB, 80.02 KB, 100.01 KB,
> 116.01 KB, 120.01 KB, 140.02 KB, 152.02 KB, 156.01 KB, 180.01 KB, 204.01
> KB, 232.05 KB, 252.01 KB, 268.02 KB, 284.01 KB, 296.01 KB, 316.01 KB,
> 332.01 KB, 344.01 KB, 364.01 KB, 368.01 KB, 404.01 KB, 436.01 KB, 440.01
> KB, 444.01 KB, 472.01 KB, 484.01 KB, 496.01 KB, 516.01 KB, 528.01 KB,
> 548.01 KB, 556.01 KB, 564.02 KB, 572.01 KB
>          - AverageThreadTokens: 0.00
>          - BloomFilterBytes: 0
>          - PeakMemoryUsage: 580.02 KB (593936)
>          - PeakReservation: 0
>          - PeakUsedReservation: 0
>          - PerHostPeakMemUsage: 144.57 MB (151595498)
>          - RowsProduced: 286 (286)
>          - TotalNetworkReceiveTime: 16s926ms
>          - TotalNetworkSendTime: 0.000ns
>          - TotalStorageWaitTime: 0.000ns
>          - TotalThreadsInvoluntaryContextSwitches: 0 (0)
>          - TotalThreadsTotalWallClockTime: 16s929ms
>            - TotalThreadsSysTime: 849.000us
>            - TotalThreadsUserTime: 3.925ms
>          - TotalThreadsVoluntaryContextSwitches: 144 (144)
>         Fragment Instance Lifecycle Timings:
>            - ExecTime: 16s614ms
>              - ExecTreeExecTime: 16s612ms
>            - OpenTime: 314.632ms
>              - ExecTreeOpenTime: 314.614ms
>            - PrepareTime: 26.668ms
>              - ExecTreePrepareTime: 43.413us
>         PLAN_ROOT_SINK:
>            - PeakMemoryUsage: 0
>         CodeGen:(Total: 26.326ms, non-child: 26.326ms, % non-child:
> 100.00%)
>            - CodegenTime: 0.000ns
>            - CompileTime: 0.000ns
>            - LoadTime: 0.000ns
>            - ModuleBitcodeSize: 1.86 MB (1950884)
>            - NumFunctions: 0 (0)
>            - NumInstructions: 0 (0)
>            - OptimizationTime: 0.000ns
>            - PeakMemoryUsage: 0
>            - PrepareTime: 25.464ms
>         EXCHANGE_NODE (id=1):(Total: 16s927ms, non-child: 16s927ms, %
> non-child: 100.00%)
>            - ConvertRowBatchTime: 290.044us
>            - PeakMemoryUsage: 0
>            - RowsReturned: 286 (286)
>            - RowsReturnedRate: 16.00 /sec
>           DataStreamReceiver:
>             BytesReceived(500.000ms): 0, 3.67 KB, 7.30 KB, 9.67 KB, 10.76
> KB, 11.00 KB, 14.20 KB, 16.04 KB, 16.27 KB, 19.18 KB, 22.14 KB, 25.12 KB,
> 26.80 KB, 29.04 KB, 30.00 KB, 30.68 KB, 32.34 KB, 34.66 KB, 35.48 KB, 36.65
> KB, 36.91 KB, 39.72 KB, 42.90 KB, 43.14 KB, 43.39 KB, 45.58 KB, 46.29 KB,
> 47.24 KB, 48.57 KB, 50.38 KB, 51.53 KB, 52.24 KB, 52.95 KB, 53.84 KB
>              - BytesReceived: 54.30 KB (55602)
>              - DeserializeRowBatchTimer: 2.272ms
>              - FirstBatchArrivalWaitTime: 314.602ms
>              - PeakMemoryUsage: 4.09 KB (4184)
>              - SendersBlockedTimer: 0.000ns
>              - SendersBlockedTotalTimer(*): 0.000ns
>     Averaged Fragment F00:(Total: 15s383ms, non-child: 436.851ms, %
> non-child: 2.84%)
>       split sizes:  min: 2.12 GB, max: 2.13 GB, avg: 2.12 GB, stddev: 5.95
> MB
>       completion times: min:13s654ms  max:16s956ms  mean: 15s385ms
> stddev:1s352ms
>       execution rates: min:128.65 MB/sec  max:158.83 MB/sec  mean:142.33
> MB/sec  stddev:12.48 MB/sec
>       num instances: 3
>        - AverageThreadTokens: 1.99
>        - BloomFilterBytes: 0
>        - PeakMemoryUsage: 144.09 MB (151090554)
>        - PeakReservation: 0
>        - PeakUsedReservation: 0
>        - PerHostPeakMemUsage: 144.25 MB (151259858)
>        - RowsProduced: 95 (95)
>        - TotalNetworkReceiveTime: 0.000ns
>        - TotalNetworkSendTime: 21.262us
>        - TotalStorageWaitTime: 456.615ms
>        - TotalThreadsInvoluntaryContextSwitches: 50 (50)
>        - TotalThreadsTotalWallClockTime: 30s671ms
>          - TotalThreadsSysTime: 1s040ms
>          - TotalThreadsUserTime: 14s297ms
>        - TotalThreadsVoluntaryContextSwitches: 1.22K (1222)
>       Fragment Instance Lifecycle Timings:
>          - ExecTime: 15s314ms
>            - ExecTreeExecTime: 14s876ms
>          - OpenTime: 43.132ms
>            - ExecTreeOpenTime: 44.527us
>          - PrepareTime: 26.101ms
>            - ExecTreePrepareTime: 214.813us
>       DataStreamSender (dst_id=1):(Total: 3.177ms, non-child: 3.177ms, %
> non-child: 100.00%)
>          - BytesSent: 18.10 KB (18534)
>          - NetworkThroughput(*): 9.20 MB/sec
>          - OverallThroughput: 5.57 MB/sec
>          - PeakMemoryUsage: 456.00 B (456)
>          - RowsReturned: 95 (95)
>          - SerializeBatchTime: 1.049ms
>          - TransmitDataRPCTime: 1.944ms
>          - UncompressedRowBatchSize: 38.43 KB (39355)
>       CodeGen:(Total: 67.190ms, non-child: 67.190ms, % non-child: 100.00%)
>          - CodegenTime: 1.256ms
>          - CompileTime: 10.469ms
>          - LoadTime: 0.000ns
>          - ModuleBitcodeSize: 1.86 MB (1950884)
>          - NumFunctions: 32 (32)
>          - NumInstructions: 483 (483)
>          - OptimizationTime: 31.695ms
>          - PeakMemoryUsage: 241.50 KB (247296)
>          - PrepareTime: 24.214ms
>       HDFS_SCAN_NODE (id=0):(Total: 14s876ms, non-child: 14s876ms, %
> non-child: 100.00%)
>          - AverageHdfsReadThreadConcurrency: 0.13
>          - AverageScannerThreadConcurrency: 1.00
>          - BytesRead: 2.04 GB (2192368083)
>          - BytesReadDataNodeCache: 0
>          - BytesReadLocal: 2.04 GB (2192368083)
>          - BytesReadRemoteUnexpected: 0
>          - BytesReadShortCircuit: 2.04 GB (2192368083)
>          - CachedFileHandlesHitCount: 0 (0)
>          - CachedFileHandlesMissCount: 165 (165)
>          - CollectionItemsRead: 0 (0)
>          - DecompressionTime: 3s516ms
>          - MaxCompressedTextFileLength: 0
>          - NumColumns: 18 (18)
>          - NumDictFilteredRowGroups: 0 (0)
>          - NumDisksAccessed: 4 (4)
>          - NumRowGroups: 9 (9)
>          - NumScannerThreadsStarted: 1 (1)
>          - NumScannersWithNoReads: 0 (0)
>          - NumStatsFilteredRowGroups: 0 (0)
>          - PeakMemoryUsage: 143.22 MB (150173543)
>          - PerReadThreadRawHdfsThroughput: 1.58 GB/sec
>          - RemoteScanRanges: 0 (0)
>          - RowBatchQueueGetWaitTime: 14s871ms
>          - RowBatchQueuePutWaitTime: 0.000ns
>          - RowsRead: 49.70M (49702305)
>          - RowsReturned: 95 (95)
>          - RowsReturnedRate: 5.00 /sec
>          - ScanRangesComplete: 9 (9)
>          - ScannerThreadsInvoluntaryContextSwitches: 41 (41)
>          - ScannerThreadsTotalWallClockTime: 15s313ms
>            - MaterializeTupleTime(*): 11s093ms
>            - ScannerThreadsSysTime: 645.800ms
>            - ScannerThreadsUserTime: 14s202ms
>          - ScannerThreadsVoluntaryContextSwitches: 904 (904)
>          - TotalRawHdfsReadTime(*): 1s293ms
>          - TotalReadThroughput: 136.47 MB/sec
>     Fragment F00:
>       Instance b4f4f6493e9fe5a:fc9961d600000002
> (host=bigdb02:22000):(Total: 16s955ms, non-child: 435.525ms, % non-child:
> 2.57%)
>         Hdfs split stats (<volume id>:<# splits>/<split lengths>):
> 1:1/162.67 MB 0:1/252.38 MB 2:2/504.68 MB 4:3/756.96 MB 3:2/504.74 MB
>         MemoryUsage(500.000ms): 8.45 KB, 133.20 MB, 121.00 MB, 109.06 MB,
> 132.29 MB, 133.13 MB, 122.50 MB, 104.69 MB, 128.66 MB, 132.51 MB, 121.07
> MB, 110.50 MB, 86.04 MB, 97.95 MB, 82.56 MB, 121.29 MB, 133.22 MB, 126.65
> MB, 111.19 MB, 130.34 MB, 131.22 MB, 113.42 MB, 103.79 MB, 131.90 MB,
> 122.21 MB, 114.54 MB, 94.17 MB, 132.95 MB, 126.18 MB, 110.78 MB, 12.02 MB,
> 130.73 MB, 121.57 MB, 112.30 MB
>         ThreadUsage(500.000ms): 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
> 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
>          - AverageThreadTokens: 1.97
>          - BloomFilterBytes: 0
>          - PeakMemoryUsage: 144.09 MB (151092462)
>          - PeakReservation: 0
>          - PeakUsedReservation: 0
>          - PerHostPeakMemUsage: 144.09 MB (151092462)
>          - RowsProduced: 93 (93)
>          - TotalNetworkReceiveTime: 0.000ns
>          - TotalNetworkSendTime: 24.406us
>          - TotalStorageWaitTime: 489.566ms
>          - TotalThreadsInvoluntaryContextSwitches: 37 (37)
>          - TotalThreadsTotalWallClockTime: 33s813ms
>            - TotalThreadsSysTime: 1s066ms
>            - TotalThreadsUserTime: 15s812ms
>          - TotalThreadsVoluntaryContextSwitches: 1.38K (1383)
>         Fragment Instance Lifecycle Timings:
>            - ExecTime: 16s884ms
>              - ExecTreeExecTime: 16s448ms
>            - OpenTime: 44.632ms
>              - ExecTreeOpenTime: 47.029us
>            - PrepareTime: 25.728ms
>              - ExecTreePrepareTime: 176.442us
>         DataStreamSender (dst_id=1):(Total: 3.196ms, non-child: 3.196ms, %
> non-child: 100.00%)
>            - BytesSent: 18.56 KB (19004)
>            - NetworkThroughput(*): 9.12 MB/sec
>            - OverallThroughput: 5.67 MB/sec
>            - PeakMemoryUsage: 456.00 B (456)
>            - RowsReturned: 93 (93)
>            - SerializeBatchTime: 1.026ms
>            - TransmitDataRPCTime: 1.988ms
>            - UncompressedRowBatchSize: 37.12 KB (38010)
>         CodeGen:(Total: 68.333ms, non-child: 68.333ms, % non-child:
> 100.00%)
>            - CodegenTime: 1.251ms
>            - CompileTime: 10.833ms
>            - LoadTime: 0.000ns
>            - ModuleBitcodeSize: 1.86 MB (1950884)
>            - NumFunctions: 32 (32)
>            - NumInstructions: 483 (483)
>            - OptimizationTime: 32.844ms
>            - PeakMemoryUsage: 241.50 KB (247296)
>            - PrepareTime: 23.839ms
>         HDFS_SCAN_NODE (id=0):(Total: 16s448ms, non-child: 16s448ms, %
> non-child: 100.00%)
>           Hdfs split stats (<volume id>:<# splits>/<split lengths>):
> 1:1/162.67 MB 0:1/252.38 MB 2:2/504.68 MB 4:3/756.96 MB 3:2/504.74 MB
>           ExecOption: PARQUET Codegen Enabled, Codegen enabled: 9 out of 9
>           Hdfs Read Thread Concurrency Bucket: 0:81.82% 1:18.18% 2:0% 3:0%
> 4:0% 5:0% 6:0% 7:0% 8:0% 9:0% 10:0% 11:0%
>           File Formats: PARQUET/SNAPPY:9
>           BytesRead(500.000ms): 0, 134.13 MB, 186.80 MB, 234.80 MB, 354.64
> MB, 402.64 MB, 447.06 MB, 495.06 MB, 614.97 MB, 646.97 MB, 691.70 MB,
> 731.70 MB, 757.29 MB, 866.45 MB, 906.45 MB, 1007.14 MB, 1.05 GB, 1.09 GB,
> 1.14 GB, 1.25 GB, 1.30 GB, 1.34 GB, 1.39 GB, 1.51 GB, 1.56 GB, 1.60 GB,
> 1.64 GB, 1.76 GB, 1.81 GB, 1.86 GB, 1.89 GB, 2.02 GB, 2.07 GB, 2.11 GB
>            - FooterProcessingTime: (Avg: 3.282ms ; Min: 1.994ms ; Max:
> 10.196ms ; Number of samples: 9)
>            - AverageHdfsReadThreadConcurrency: 0.18
>            - AverageScannerThreadConcurrency: 1.00
>            - BytesRead: 2.13 GB (2288275802)
>            - BytesReadDataNodeCache: 0
>            - BytesReadLocal: 2.13 GB (2288275802)
>            - BytesReadRemoteUnexpected: 0
>            - BytesReadShortCircuit: 2.13 GB (2288275802)
>            - CachedFileHandlesHitCount: 0 (0)
>            - CachedFileHandlesMissCount: 171 (171)
>            - CollectionItemsRead: 0 (0)
>            - DecompressionTime: 3s738ms
>            - MaxCompressedTextFileLength: 0
>            - NumColumns: 18 (18)
>            - NumDictFilteredRowGroups: 0 (0)
>            - NumDisksAccessed: 5 (5)
>            - NumRowGroups: 9 (9)
>            - NumScannerThreadsStarted: 1 (1)
>            - NumScannersWithNoReads: 0 (0)
>            - NumStatsFilteredRowGroups: 0 (0)
>            - PeakMemoryUsage: 143.25 MB (150207517)
>            - PerReadThreadRawHdfsThroughput: 1.50 GB/sec
>            - RemoteScanRanges: 0 (0)
>            - RowBatchQueueGetWaitTime: 16s443ms
>            - RowBatchQueuePutWaitTime: 0.000ns
>            - RowsRead: 51.90M (51903646)
>            - RowsReturned: 93 (93)
>            - RowsReturnedRate: 5.00 /sec
>            - ScanRangesComplete: 9 (9)
>            - ScannerThreadsInvoluntaryContextSwitches: 26 (26)
>            - ScannerThreadsTotalWallClockTime: 16s884ms
>              - MaterializeTupleTime(*): 12s402ms
>              - ScannerThreadsSysTime: 667.204ms
>              - ScannerThreadsUserTime: 15s717ms
>            - ScannerThreadsVoluntaryContextSwitches: 1.05K (1054)
>            - TotalRawHdfsReadTime(*): 1s422ms
>            - TotalReadThroughput: 127.33 MB/sec
>       Instance b4f4f6493e9fe5a:fc9961d600000001
> (host=bigdb01:22000):(Total: 15s542ms, non-child: 413.267ms, % non-child:
> 2.66%)
>         Hdfs split stats (<volume id>:<# splits>/<split lengths>):
> 0:1/252.36 MB 3:1/149.88 MB 2:2/504.73 MB 4:1/252.35 MB 1:4/1009.49 MB
>         MemoryUsage(500.000ms): 134.10 MB, 133.28 MB, 126.63 MB, 110.79
> MB, 88.39 MB, 134.59 MB, 126.31 MB, 114.98 MB, 91.04 MB, 131.00 MB, 127.24
> MB, 120.06 MB, 90.81 MB, 97.26 MB, 83.69 MB, 129.41 MB, 128.89 MB, 123.06
> MB, 111.92 MB, 89.77 MB, 130.64 MB, 122.12 MB, 106.75 MB, 130.91 MB, 128.53
> MB, 127.94 MB, 110.48 MB, 40.90 MB, 129.82 MB, 122.63 MB, 104.94 MB
>         ThreadUsage(500.000ms): 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
> 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
>          - AverageThreadTokens: 2.00
>          - BloomFilterBytes: 0
>          - PeakMemoryUsage: 144.09 MB (151087586)
>          - PeakReservation: 0
>          - PeakUsedReservation: 0
>          - PerHostPeakMemUsage: 144.57 MB (151595498)
>          - RowsProduced: 89 (89)
>          - TotalNetworkReceiveTime: 0.000ns
>          - TotalNetworkSendTime: 17.632us
>          - TotalStorageWaitTime: 434.756ms
>          - TotalThreadsInvoluntaryContextSwitches: 95 (95)
>          - TotalThreadsTotalWallClockTime: 30s991ms
>            - TotalThreadsSysTime: 1s050ms
>            - TotalThreadsUserTime: 14s447ms
>          - TotalThreadsVoluntaryContextSwitches: 874 (874)
>         Fragment Instance Lifecycle Timings:
>            - ExecTime: 15s475ms
>              - ExecTreeExecTime: 15s061ms
>            - OpenTime: 40.572ms
>              - ExecTreeOpenTime: 41.278us
>            - PrepareTime: 26.145ms
>              - ExecTreePrepareTime: 222.311us
>         DataStreamSender (dst_id=1):(Total: 3.198ms, non-child: 3.198ms, %
> non-child: 100.00%)
>            - BytesSent: 16.21 KB (16600)
>            - NetworkThroughput(*): 7.50 MB/sec
>            - OverallThroughput: 4.95 MB/sec
>            - PeakMemoryUsage: 456.00 B (456)
>            - RowsReturned: 89 (89)
>            - SerializeBatchTime: 856.396us
>            - TransmitDataRPCTime: 2.112ms
>            - UncompressedRowBatchSize: 36.08 KB (36942)
>         CodeGen:(Total: 64.650ms, non-child: 64.650ms, % non-child:
> 100.00%)
>            - CodegenTime: 1.228ms
>            - CompileTime: 9.773ms
>            - LoadTime: 0.000ns
>            - ModuleBitcodeSize: 1.86 MB (1950884)
>            - NumFunctions: 32 (32)
>            - NumInstructions: 483 (483)
>            - OptimizationTime: 29.748ms
>            - PeakMemoryUsage: 241.50 KB (247296)
>            - PrepareTime: 24.331ms
>         HDFS_SCAN_NODE (id=0):(Total: 15s061ms, non-child: 15s061ms, %
> non-child: 100.00%)
>           Hdfs split stats (<volume id>:<# splits>/<split lengths>):
> 0:1/252.36 MB 3:1/149.88 MB 2:2/504.73 MB 4:1/252.35 MB 1:4/1009.49 MB
>           ExecOption: PARQUET Codegen Enabled, Codegen enabled: 9 out of 9
>           Hdfs Read Thread Concurrency Bucket: 0:87.1% 1:12.9% 2:0% 3:0%
> 4:0% 5:0% 6:0% 7:0% 8:0% 9:0% 10:0% 11:0%
>           File Formats: PARQUET/SNAPPY:8 PARQUET/(NONE,SNAPPY):1
>           BytesRead(500.000ms): 102.00 MB, 150.00 MB, 194.70 MB, 226.70
> MB, 252.45 MB, 378.45 MB, 423.23 MB, 463.23 MB, 504.89 MB, 623.15 MB,
> 667.91 MB, 715.91 MB, 757.35 MB, 854.18 MB, 894.18 MB, 1009.41 MB, 1.02 GB,
> 1.07 GB, 1.11 GB, 1.13 GB, 1.26 GB, 1.32 GB, 1.36 GB, 1.48 GB, 1.52 GB,
> 1.56 GB, 1.60 GB, 1.65 GB, 1.77 GB, 1.82 GB, 1.87 GB
>            - FooterProcessingTime: (Avg: 3.554ms ; Min: 1.844ms ; Max:
> 13.695ms ; Number of samples: 9)
>            - AverageHdfsReadThreadConcurrency: 0.13
>            - AverageScannerThreadConcurrency: 1.00
>            - BytesRead: 1.88 GB (2013820022)
>            - BytesReadDataNodeCache: 0
>            - BytesReadLocal: 1.88 GB (2013820022)
>            - BytesReadRemoteUnexpected: 0
>            - BytesReadShortCircuit: 1.88 GB (2013820022)
>            - CachedFileHandlesHitCount: 0 (0)
>            - CachedFileHandlesMissCount: 154 (154)
>            - CollectionItemsRead: 0 (0)
>            - DecompressionTime: 3s506ms
>            - MaxCompressedTextFileLength: 0
>            - NumColumns: 18 (18)
>            - NumDictFilteredRowGroups: 1 (1)
>            - NumDisksAccessed: 5 (5)
>            - NumRowGroups: 9 (9)
>            - NumScannerThreadsStarted: 1 (1)
>            - NumScannersWithNoReads: 0 (0)
>            - NumStatsFilteredRowGroups: 0 (0)
>            - PeakMemoryUsage: 143.15 MB (150100301)
>            - PerReadThreadRawHdfsThroughput: 1.56 GB/sec
>            - RemoteScanRanges: 0 (0)
>            - RowBatchQueueGetWaitTime: 15s057ms
>            - RowBatchQueuePutWaitTime: 0.000ns
>            - RowsRead: 45.60M (45604014)
>            - RowsReturned: 89 (89)
>            - RowsReturnedRate: 5.00 /sec
>            - ScanRangesComplete: 9 (9)
>            - ScannerThreadsInvoluntaryContextSwitches: 82 (82)
>            - ScannerThreadsTotalWallClockTime: 15s475ms
>              - MaterializeTupleTime(*): 11s276ms
>              - ScannerThreadsSysTime: 674.148ms
>              - ScannerThreadsUserTime: 14s361ms
>            - ScannerThreadsVoluntaryContextSwitches: 567 (567)
>            - TotalRawHdfsReadTime(*): 1s202ms
>            - TotalReadThroughput: 123.27 MB/sec
>       Instance b4f4f6493e9fe5a:fc9961d600000003
> (host=bigdb03:22000):(Total: 13s653ms, non-child: 461.761ms, % non-child:
> 3.38%)
>         Hdfs split stats (<volume id>:<# splits>/<split lengths>):
> 2:2/504.66 MB 0:1/252.37 MB 3:3/654.56 MB 1:3/757.17 MB
>         MemoryUsage(500.000ms): 132.79 MB, 123.57 MB, 108.74 MB, 97.90 MB,
> 79.75 MB, 129.78 MB, 125.51 MB, 110.49 MB, 69.49 MB, 141.21 MB, 114.37 MB,
> 89.40 MB, 130.39 MB, 121.69 MB, 92.30 MB, 135.26 MB, 127.93 MB, 104.04 MB,
> 130.38 MB, 120.88 MB, 104.35 MB, 132.65 MB, 123.71 MB, 106.08 MB, 130.69
> MB, 123.63 MB, 112.19 MB
>         ThreadUsage(500.000ms): 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
> 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
>          - AverageThreadTokens: 2.00
>          - BloomFilterBytes: 0
>          - PeakMemoryUsage: 144.09 MB (151091614)
>          - PeakReservation: 0
>          - PeakUsedReservation: 0
>          - PerHostPeakMemUsage: 144.09 MB (151091614)
>          - RowsProduced: 104 (104)
>          - TotalNetworkReceiveTime: 0.000ns
>          - TotalNetworkSendTime: 21.748us
>          - TotalStorageWaitTime: 445.522ms
>          - TotalThreadsInvoluntaryContextSwitches: 20 (20)
>          - TotalThreadsTotalWallClockTime: 27s208ms
>            - TotalThreadsSysTime: 1s005ms
>            - TotalThreadsUserTime: 12s631ms
>          - TotalThreadsVoluntaryContextSwitches: 1.41K (1410)
>         Fragment Instance Lifecycle Timings:
>            - ExecTime: 13s582ms
>              - ExecTreeExecTime: 13s119ms
>            - OpenTime: 44.193ms
>              - ExecTreeOpenTime: 45.276us
>            - PrepareTime: 26.432ms
>              - ExecTreePrepareTime: 245.688us
>         DataStreamSender (dst_id=1):(Total: 3.138ms, non-child: 3.138ms, %
> non-child: 100.00%)
>            - BytesSent: 19.53 KB (19998)
>            - NetworkThroughput(*): 11.00 MB/sec
>            - OverallThroughput: 6.08 MB/sec
>            - PeakMemoryUsage: 456.00 B (456)
>            - RowsReturned: 104 (104)
>            - SerializeBatchTime: 1.264ms
>            - TransmitDataRPCTime: 1.733ms
>            - UncompressedRowBatchSize: 42.10 KB (43114)
>         CodeGen:(Total: 68.585ms, non-child: 68.585ms, % non-child:
> 100.00%)
>            - CodegenTime: 1.290ms
>            - CompileTime: 10.802ms
>            - LoadTime: 0.000ns
>            - ModuleBitcodeSize: 1.86 MB (1950884)
>            - NumFunctions: 32 (32)
>            - NumInstructions: 483 (483)
>            - OptimizationTime: 32.495ms
>            - PeakMemoryUsage: 241.50 KB (247296)
>            - PrepareTime: 24.471ms
>         HDFS_SCAN_NODE (id=0):(Total: 13s119ms, non-child: 13s119ms, %
> non-child: 100.00%)
>           Hdfs split stats (<volume id>:<# splits>/<split lengths>):
> 2:2/504.66 MB 0:1/252.37 MB 3:3/654.56 MB 1:3/757.17 MB
>           ExecOption: PARQUET Codegen Enabled, Codegen enabled: 9 out of 9
>           Hdfs Read Thread Concurrency Bucket: 0:92.59% 1:7.407% 2:0% 3:0%
> 4:0% 5:0% 6:0% 7:0% 8:0% 9:0% 10:0% 11:0%
>           File Formats: PARQUET/SNAPPY:9
>           BytesRead(500.000ms): 102.13 MB, 170.89 MB, 226.89 MB, 333.23
> MB, 389.23 MB, 512.37 MB, 565.15 MB, 621.15 MB, 696.90 MB, 804.95 MB,
> 865.63 MB, 907.22 MB, 1.02 GB, 1.08 GB, 1.13 GB, 1.26 GB, 1.32 GB, 1.38 GB,
> 1.50 GB, 1.56 GB, 1.62 GB, 1.74 GB, 1.80 GB, 1.85 GB, 1.99 GB, 2.04 GB,
> 2.09 GB
>            - FooterProcessingTime: (Avg: 4.124ms ; Min: 1.805ms ; Max:
> 19.516ms ; Number of samples: 9)
>            - AverageHdfsReadThreadConcurrency: 0.07
>            - AverageScannerThreadConcurrency: 1.00
>            - BytesRead: 2.12 GB (2275008425)
>            - BytesReadDataNodeCache: 0
>            - BytesReadLocal: 2.12 GB (2275008425)
>            - BytesReadRemoteUnexpected: 0
>            - BytesReadShortCircuit: 2.12 GB (2275008425)
>            - CachedFileHandlesHitCount: 0 (0)
>            - CachedFileHandlesMissCount: 171 (171)
>            - CollectionItemsRead: 0 (0)
>            - DecompressionTime: 3s303ms
>            - MaxCompressedTextFileLength: 0
>            - NumColumns: 18 (18)
>            - NumDictFilteredRowGroups: 0 (0)
>            - NumDisksAccessed: 4 (4)
>            - NumRowGroups: 9 (9)
>            - NumScannerThreadsStarted: 1 (1)
>            - NumScannersWithNoReads: 0 (0)
>            - NumStatsFilteredRowGroups: 0 (0)
>            - PeakMemoryUsage: 143.25 MB (150212813)
>            - PerReadThreadRawHdfsThroughput: 1.69 GB/sec
>            - RemoteScanRanges: 0 (0)
>            - RowBatchQueueGetWaitTime: 13s114ms
>            - RowBatchQueuePutWaitTime: 0.000ns
>            - RowsRead: 51.60M (51599256)
>            - RowsReturned: 104 (104)
>            - RowsReturnedRate: 7.00 /sec
>            - ScanRangesComplete: 9 (9)
>            - ScannerThreadsInvoluntaryContextSwitches: 17 (17)
>            - ScannerThreadsTotalWallClockTime: 13s581ms
>              - MaterializeTupleTime(*): 9s602ms
>              - ScannerThreadsSysTime: 596.050ms
>              - ScannerThreadsUserTime: 12s527ms
>            - ScannerThreadsVoluntaryContextSwitches: 1.09K (1092)
>            - TotalRawHdfsReadTime(*): 1s255ms
>            - TotalReadThroughput: 158.81 MB/sec
>
> On 08/03/2018 02:22 AM, Tim Armstrong wrote:
>
> Can you include a complete profile?
>
> On Thu, Aug 2, 2018 at 12:31 AM, meijie <me...@yandex.com> wrote:
>
>> Hi, Every One
>>
>> I found the sql (select * from parquet_session_1 where browser = 'Firefox
>> 20') was very slow at my impala cluster. which needed 15 seconds to
>> executed and NumScannerThreadsStarted is 1 , it wasn't as expected.
>>
>> the part profile as blew:
>>
>>            - FooterProcessingTime: (Avg: 4.430ms ; Min: 2.924ms ; Max:
>> 9.559ms ; Number of samples: 9)
>>            - AverageHdfsReadThreadConcurrency: 0.00
>>            - AverageScannerThreadConcurrency: 1.00
>>            - BytesRead: 2.13 GB (2288275802)
>>            - BytesReadDataNodeCache: 0
>>            - BytesReadLocal: 2.13 GB (2288275802)
>>            - BytesReadRemoteUnexpected: 0
>>            - BytesReadShortCircuit: 2.13 GB (2288275802)
>>            - CachedFileHandlesHitCount: 0 (0)
>>            - CachedFileHandlesMissCount: 171 (171)
>>            - CollectionItemsRead: 0 (0)
>>            - DecompressionTime: 3s264ms
>>            - MaxCompressedTextFileLength: 0
>>            - NumColumns: 18 (18)
>>            - NumDictFilteredRowGroups: 0 (0)
>>            - NumDisksAccessed: 5 (5)
>>            - NumRowGroups: 9 (9)
>>            - NumScannerThreadsStarted: 1 (1)
>>            - NumScannersWithNoReads: 0 (0)
>>            - NumStatsFilteredRowGroups: 0 (0)
>>            - PeakMemoryUsage: 143.25 MB (user150207517)
>>            - PerReadThreadRawHdfsThroughput: 1.30 GB/sec
>>            - RemoteScanRanges: 0 (0)
>>            - RowBatchQueueGetWaitTime: 14s216ms
>>            - RowBatchQueuePutWaitTime: 0.000ns
>>            - RowsRead: 51.90M (51903646)
>>            - RowsReturned: 93 (93)
>>            - RowsReturnedRate: 6.00 /sec
>>            - ScanRangesComplete: 9 (9)
>>            - ScannerThreadsInvoluntaryContextSwitches: 18 (18)
>>            - ScannerThreadsTotalWallClockTime: 14s678ms
>>              - MaterializeTupleTime(*): 10s639ms
>>              - ScannerThreadsSysTime: 698.325ms
>>              - ScannerThreadsUserTime: 13s452ms
>>            - ScannerThreadsVoluntaryContextSwitches: 1.38K (1382)
>>            - TotalRawHdfsReadTime(*): 1s637ms
>>            - TotalReadThroughput: 148.73 MB/sec
>>
>> But, if I only select one column(select uuid from parquet_session_1 where
>> browser = 'Firefox 20'), It just needed 0.5 seconds and
>> NumScannerThreadsStarted is 4. I don't know what affects the number of scan
>> threads.
>>
>> Impala Version: 2.11.0-cdh5.14.0 RELEASE, the sql submitted by
>> impala-shell
>>
>> Thanks
>>
>> --
>> Best Regard
>> meijie
>>
>>
>
> --
> Best Regard
> meijie
>
>

Re: Slow sql because NumScannerThreadsStarted is 1

Posted by meijie <me...@yandex.com>.
Thank you for your response. blew is complete profile

Query Runtime Profile:
Query (id=b4f4f6493e9fe5a:fc9961d600000000):
   Summary:
     Session ID: ef47d3119c267424:d9df967cefd5a883
     Session Type: BEESWAX
     Start Time: 2018-08-03 09:38:45.127171000
     End Time: 2018-08-03 09:39:02.285440000
     Query Type: QUERY
     Query State: FINISHED
     Query Status: OK
     Impala Version: impalad version 2.11.0-cdh5.14.0 RELEASE (build 
d68206561bce6b26762d62c01a78e6cd27aa7690)
     User: root
     Connected User: root
     Delegated User:
     Network Address: ::ffff:172.16.20.50:55138
     Default Db: bigdb_press_test
     Sql Statement: select * from parquet_session_1 where browser = 
'Firefox 20'
     Coordinator: bigdb01:22000
     Query Options (set by configuration): MEM_LIMIT=419430400
     Query Options (set by configuration and planner): 
MEM_LIMIT=419430400,MT_DOP=0
     Plan:
----------------
Max Per-Host Resource Reservation: Memory=0B
Per-Host Resource Estimates: Memory=1.55GB

F01:PLAN FRAGMENT [UNPARTITIONED] hosts=1 instances=1
|  Per-Host Resources: mem-estimate=0B mem-reservation=0B
PLAN-ROOT SINK
|  mem-estimate=0B mem-reservation=0B
|
01:EXCHANGE [UNPARTITIONED]
|  mem-estimate=0B mem-reservation=0B
|  tuple-ids=0 row-size=396B cardinality=1140504
|
F00:PLAN FRAGMENT [RANDOM] hosts=3 instances=3
Per-Host Resources: mem-estimate=1.55GB mem-reservation=0B
00:SCAN HDFS [bigdb_press_test.parquet_session_1, RANDOM]
    partitions=1/1 files=27 size=6.37GB
    predicates: browser = 'Firefox 20'
    stats-rows=155108551 extrapolated-rows=disabled
    table stats: rows=155108551 size=6.37GB
    column stats: all
    parquet statistics predicates: browser = 'Firefox 20'
    parquet dictionary predicates: browser = 'Firefox 20'
    mem-estimate=1.55GB mem-reservation=0B
    tuple-ids=0 row-size=396B cardinality=1140504
----------------
     Estimated Per-Host Mem: 1660944384
     Per Host Min Reservation: bigdb01:22000(0) bigdb02:22000(0) 
bigdb03:22000(0)
     Request Pool: root.default
     Admission result: Admitted immediately
     ExecSummary:
Operator       #Hosts  Avg Time  Max Time  #Rows  Est. #Rows Peak Mem  
Est. Peak Mem  Detail
---------------------------------------------------------------------------------------------------------------------------
01:EXCHANGE         1  16s927ms  16s927ms    286 1.14M          
0              0 UNPARTITIONED
00:SCAN HDFS        3  14s876ms  16s448ms    286       1.14M 143.25 
MB        1.55 GB  bigdb_press_test.parquet_session_1
     Errors:
     Planner Timeline: 4.874ms
        - Analysis finished: 1.479ms (1.479ms)
        - Value transfer graph computed: 1.694ms (215.699us)
        - Single node plan created: 3.140ms (1.445ms)
        - Runtime filters computed: 3.256ms (116.502us)
        - Distributed plan created: 3.324ms (67.434us)
        - Lineage info computed: 3.467ms (143.587us)
        - Planning finished: 4.874ms (1.406ms)
     Query Timeline: 17s161ms
        - Query submitted: 75.337us (75.337us)
        - Planning finished: 6.136ms (6.061ms)
        - Submit for admission: 6.636ms (500.206us)
        - Completed admission: 6.716ms (79.190us)
        - Ready to start on 3 backends: 6.935ms (219.689us)
        - All 3 execution backends (4 fragment instances) started: 
8.842ms (1.906ms)
        - Rows available: 351.212ms (342.370ms)
        - First row fetched: 441.198ms (89.985ms)
        - Unregister query: 17s158ms (16s717ms)
      - ComputeScanRangeAssignmentTimer: 109.935us
   ImpalaServer:
      - ClientFetchWaitTimer: 280.979ms
      - RowMaterializationTimer: 16s526ms
   Execution Profile b4f4f6493e9fe5a:fc9961d600000000:(Total: 16s870ms, 
non-child: 0.000ns, % non-child: 0.00%)
     Number of filters: 0
     Filter routing table:
  ID  Src. Node  Tgt. Node(s)  Target type  Partition filter Pending 
(Expected)  First arrived  Completed   Enabled
-------------------------------------------------------------------------------------------------------------------

     Backend startup latencies: Count: 3, min / max: 1ms / 2ms, 25th 
%-ile: 1ms, 50th %-ile: 2ms, 75th %-ile: 2ms, 90th %-ile: 2ms, 95th 
%-ile: 2ms, 99.9th %-ile: 2ms
     Per Node Peak Memory Usage: bigdb01:22000(144.57 MB) 
bigdb02:22000(144.09 MB) bigdb03:22000(144.09 MB)
      - FiltersReceived: 0 (0)
      - FinalizationTimer: 0.000ns
      - NumBackends: 3 (3)
      - NumFragmentInstances: 4 (4)
      - NumFragments: 2 (2)
     Averaged Fragment F01:(Total: 16s955ms, non-child: 2.075ms, % 
non-child: 0.01%)
       split sizes:  min: 0, max: 0, avg: 0, stddev: 0
       completion times: min:16s957ms  max:16s957ms  mean: 16s957ms  
stddev:0.000ns
       execution rates: min:0.00 /sec  max:0.00 /sec  mean:0.00 /sec  
stddev:0.00 /sec
       num instances: 1
        - AverageThreadTokens: 0.00
        - BloomFilterBytes: 0
        - PeakMemoryUsage: 580.02 KB (593936)
        - PeakReservation: 0
        - PeakUsedReservation: 0
        - PerHostPeakMemUsage: 144.57 MB (151595498)
        - RowsProduced: 286 (286)
        - TotalNetworkReceiveTime: 16s926ms
        - TotalNetworkSendTime: 0.000ns
        - TotalStorageWaitTime: 0.000ns
        - TotalThreadsInvoluntaryContextSwitches: 0 (0)
        - TotalThreadsTotalWallClockTime: 16s929ms
          - TotalThreadsSysTime: 849.000us
          - TotalThreadsUserTime: 3.925ms
        - TotalThreadsVoluntaryContextSwitches: 144 (144)
       Fragment Instance Lifecycle Timings:
          - ExecTime: 16s614ms
            - ExecTreeExecTime: 16s612ms
          - OpenTime: 314.632ms
            - ExecTreeOpenTime: 314.614ms
          - PrepareTime: 26.668ms
            - ExecTreePrepareTime: 43.413us
       PLAN_ROOT_SINK:
          - PeakMemoryUsage: 0
       CodeGen:(Total: 26.326ms, non-child: 26.326ms, % non-child: 100.00%)
          - CodegenTime: 0.000ns
          - CompileTime: 0.000ns
          - LoadTime: 0.000ns
          - ModuleBitcodeSize: 1.86 MB (1950884)
          - NumFunctions: 0 (0)
          - NumInstructions: 0 (0)
          - OptimizationTime: 0.000ns
          - PeakMemoryUsage: 0
          - PrepareTime: 25.464ms
       EXCHANGE_NODE (id=1):(Total: 16s927ms, non-child: 16s927ms, % 
non-child: 100.00%)
          - ConvertRowBatchTime: 290.044us
          - PeakMemoryUsage: 0
          - RowsReturned: 286 (286)
          - RowsReturnedRate: 16.00 /sec
         DataStreamReceiver:
            - BytesReceived: 54.30 KB (55602)
            - DeserializeRowBatchTimer: 2.272ms
            - FirstBatchArrivalWaitTime: 314.602ms
            - PeakMemoryUsage: 4.09 KB (4184)
            - SendersBlockedTimer: 0.000ns
            - SendersBlockedTotalTimer(*): 0.000ns
     Coordinator Fragment F01:
       Instance b4f4f6493e9fe5a:fc9961d600000000 
(host=bigdb01:22000):(Total: 16s955ms, non-child: 2.075ms, % non-child: 
0.01%)
         MemoryUsage(500.000ms): 8.00 KB, 48.01 KB, 80.02 KB, 100.01 KB, 
116.01 KB, 120.01 KB, 140.02 KB, 152.02 KB, 156.01 KB, 180.01 KB, 204.01 
KB, 232.05 KB, 252.01 KB, 268.02 KB, 284.01 KB, 296.01 KB, 316.01 KB, 
332.01 KB, 344.01 KB, 364.01 KB, 368.01 KB, 404.01 KB, 436.01 KB, 440.01 
KB, 444.01 KB, 472.01 KB, 484.01 KB, 496.01 KB, 516.01 KB, 528.01 KB, 
548.01 KB, 556.01 KB, 564.02 KB, 572.01 KB
          - AverageThreadTokens: 0.00
          - BloomFilterBytes: 0
          - PeakMemoryUsage: 580.02 KB (593936)
          - PeakReservation: 0
          - PeakUsedReservation: 0
          - PerHostPeakMemUsage: 144.57 MB (151595498)
          - RowsProduced: 286 (286)
          - TotalNetworkReceiveTime: 16s926ms
          - TotalNetworkSendTime: 0.000ns
          - TotalStorageWaitTime: 0.000ns
          - TotalThreadsInvoluntaryContextSwitches: 0 (0)
          - TotalThreadsTotalWallClockTime: 16s929ms
            - TotalThreadsSysTime: 849.000us
            - TotalThreadsUserTime: 3.925ms
          - TotalThreadsVoluntaryContextSwitches: 144 (144)
         Fragment Instance Lifecycle Timings:
            - ExecTime: 16s614ms
              - ExecTreeExecTime: 16s612ms
            - OpenTime: 314.632ms
              - ExecTreeOpenTime: 314.614ms
            - PrepareTime: 26.668ms
              - ExecTreePrepareTime: 43.413us
         PLAN_ROOT_SINK:
            - PeakMemoryUsage: 0
         CodeGen:(Total: 26.326ms, non-child: 26.326ms, % non-child: 
100.00%)
            - CodegenTime: 0.000ns
            - CompileTime: 0.000ns
            - LoadTime: 0.000ns
            - ModuleBitcodeSize: 1.86 MB (1950884)
            - NumFunctions: 0 (0)
            - NumInstructions: 0 (0)
            - OptimizationTime: 0.000ns
            - PeakMemoryUsage: 0
            - PrepareTime: 25.464ms
         EXCHANGE_NODE (id=1):(Total: 16s927ms, non-child: 16s927ms, % 
non-child: 100.00%)
            - ConvertRowBatchTime: 290.044us
            - PeakMemoryUsage: 0
            - RowsReturned: 286 (286)
            - RowsReturnedRate: 16.00 /sec
           DataStreamReceiver:
             BytesReceived(500.000ms): 0, 3.67 KB, 7.30 KB, 9.67 KB, 
10.76 KB, 11.00 KB, 14.20 KB, 16.04 KB, 16.27 KB, 19.18 KB, 22.14 KB, 
25.12 KB, 26.80 KB, 29.04 KB, 30.00 KB, 30.68 KB, 32.34 KB, 34.66 KB, 
35.48 KB, 36.65 KB, 36.91 KB, 39.72 KB, 42.90 KB, 43.14 KB, 43.39 KB, 
45.58 KB, 46.29 KB, 47.24 KB, 48.57 KB, 50.38 KB, 51.53 KB, 52.24 KB, 
52.95 KB, 53.84 KB
              - BytesReceived: 54.30 KB (55602)
              - DeserializeRowBatchTimer: 2.272ms
              - FirstBatchArrivalWaitTime: 314.602ms
              - PeakMemoryUsage: 4.09 KB (4184)
              - SendersBlockedTimer: 0.000ns
              - SendersBlockedTotalTimer(*): 0.000ns
     Averaged Fragment F00:(Total: 15s383ms, non-child: 436.851ms, % 
non-child: 2.84%)
       split sizes:  min: 2.12 GB, max: 2.13 GB, avg: 2.12 GB, stddev: 
5.95 MB
       completion times: min:13s654ms  max:16s956ms  mean: 15s385ms  
stddev:1s352ms
       execution rates: min:128.65 MB/sec  max:158.83 MB/sec mean:142.33 
MB/sec  stddev:12.48 MB/sec
       num instances: 3
        - AverageThreadTokens: 1.99
        - BloomFilterBytes: 0
        - PeakMemoryUsage: 144.09 MB (151090554)
        - PeakReservation: 0
        - PeakUsedReservation: 0
        - PerHostPeakMemUsage: 144.25 MB (151259858)
        - RowsProduced: 95 (95)
        - TotalNetworkReceiveTime: 0.000ns
        - TotalNetworkSendTime: 21.262us
        - TotalStorageWaitTime: 456.615ms
        - TotalThreadsInvoluntaryContextSwitches: 50 (50)
        - TotalThreadsTotalWallClockTime: 30s671ms
          - TotalThreadsSysTime: 1s040ms
          - TotalThreadsUserTime: 14s297ms
        - TotalThreadsVoluntaryContextSwitches: 1.22K (1222)
       Fragment Instance Lifecycle Timings:
          - ExecTime: 15s314ms
            - ExecTreeExecTime: 14s876ms
          - OpenTime: 43.132ms
            - ExecTreeOpenTime: 44.527us
          - PrepareTime: 26.101ms
            - ExecTreePrepareTime: 214.813us
       DataStreamSender (dst_id=1):(Total: 3.177ms, non-child: 3.177ms, 
% non-child: 100.00%)
          - BytesSent: 18.10 KB (18534)
          - NetworkThroughput(*): 9.20 MB/sec
          - OverallThroughput: 5.57 MB/sec
          - PeakMemoryUsage: 456.00 B (456)
          - RowsReturned: 95 (95)
          - SerializeBatchTime: 1.049ms
          - TransmitDataRPCTime: 1.944ms
          - UncompressedRowBatchSize: 38.43 KB (39355)
       CodeGen:(Total: 67.190ms, non-child: 67.190ms, % non-child: 100.00%)
          - CodegenTime: 1.256ms
          - CompileTime: 10.469ms
          - LoadTime: 0.000ns
          - ModuleBitcodeSize: 1.86 MB (1950884)
          - NumFunctions: 32 (32)
          - NumInstructions: 483 (483)
          - OptimizationTime: 31.695ms
          - PeakMemoryUsage: 241.50 KB (247296)
          - PrepareTime: 24.214ms
       HDFS_SCAN_NODE (id=0):(Total: 14s876ms, non-child: 14s876ms, % 
non-child: 100.00%)
          - AverageHdfsReadThreadConcurrency: 0.13
          - AverageScannerThreadConcurrency: 1.00
          - BytesRead: 2.04 GB (2192368083)
          - BytesReadDataNodeCache: 0
          - BytesReadLocal: 2.04 GB (2192368083)
          - BytesReadRemoteUnexpected: 0
          - BytesReadShortCircuit: 2.04 GB (2192368083)
          - CachedFileHandlesHitCount: 0 (0)
          - CachedFileHandlesMissCount: 165 (165)
          - CollectionItemsRead: 0 (0)
          - DecompressionTime: 3s516ms
          - MaxCompressedTextFileLength: 0
          - NumColumns: 18 (18)
          - NumDictFilteredRowGroups: 0 (0)
          - NumDisksAccessed: 4 (4)
          - NumRowGroups: 9 (9)
          - NumScannerThreadsStarted: 1 (1)
          - NumScannersWithNoReads: 0 (0)
          - NumStatsFilteredRowGroups: 0 (0)
          - PeakMemoryUsage: 143.22 MB (150173543)
          - PerReadThreadRawHdfsThroughput: 1.58 GB/sec
          - RemoteScanRanges: 0 (0)
          - RowBatchQueueGetWaitTime: 14s871ms
          - RowBatchQueuePutWaitTime: 0.000ns
          - RowsRead: 49.70M (49702305)
          - RowsReturned: 95 (95)
          - RowsReturnedRate: 5.00 /sec
          - ScanRangesComplete: 9 (9)
          - ScannerThreadsInvoluntaryContextSwitches: 41 (41)
          - ScannerThreadsTotalWallClockTime: 15s313ms
            - MaterializeTupleTime(*): 11s093ms
            - ScannerThreadsSysTime: 645.800ms
            - ScannerThreadsUserTime: 14s202ms
          - ScannerThreadsVoluntaryContextSwitches: 904 (904)
          - TotalRawHdfsReadTime(*): 1s293ms
          - TotalReadThroughput: 136.47 MB/sec
     Fragment F00:
       Instance b4f4f6493e9fe5a:fc9961d600000002 
(host=bigdb02:22000):(Total: 16s955ms, non-child: 435.525ms, % 
non-child: 2.57%)
         Hdfs split stats (<volume id>:<# splits>/<split lengths>): 
1:1/162.67 MB 0:1/252.38 MB 2:2/504.68 MB 4:3/756.96 MB 3:2/504.74 MB
         MemoryUsage(500.000ms): 8.45 KB, 133.20 MB, 121.00 MB, 109.06 
MB, 132.29 MB, 133.13 MB, 122.50 MB, 104.69 MB, 128.66 MB, 132.51 MB, 
121.07 MB, 110.50 MB, 86.04 MB, 97.95 MB, 82.56 MB, 121.29 MB, 133.22 
MB, 126.65 MB, 111.19 MB, 130.34 MB, 131.22 MB, 113.42 MB, 103.79 MB, 
131.90 MB, 122.21 MB, 114.54 MB, 94.17 MB, 132.95 MB, 126.18 MB, 110.78 
MB, 12.02 MB, 130.73 MB, 121.57 MB, 112.30 MB
         ThreadUsage(500.000ms): 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
          - AverageThreadTokens: 1.97
          - BloomFilterBytes: 0
          - PeakMemoryUsage: 144.09 MB (151092462)
          - PeakReservation: 0
          - PeakUsedReservation: 0
          - PerHostPeakMemUsage: 144.09 MB (151092462)
          - RowsProduced: 93 (93)
          - TotalNetworkReceiveTime: 0.000ns
          - TotalNetworkSendTime: 24.406us
          - TotalStorageWaitTime: 489.566ms
          - TotalThreadsInvoluntaryContextSwitches: 37 (37)
          - TotalThreadsTotalWallClockTime: 33s813ms
            - TotalThreadsSysTime: 1s066ms
            - TotalThreadsUserTime: 15s812ms
          - TotalThreadsVoluntaryContextSwitches: 1.38K (1383)
         Fragment Instance Lifecycle Timings:
            - ExecTime: 16s884ms
              - ExecTreeExecTime: 16s448ms
            - OpenTime: 44.632ms
              - ExecTreeOpenTime: 47.029us
            - PrepareTime: 25.728ms
              - ExecTreePrepareTime: 176.442us
         DataStreamSender (dst_id=1):(Total: 3.196ms, non-child: 
3.196ms, % non-child: 100.00%)
            - BytesSent: 18.56 KB (19004)
            - NetworkThroughput(*): 9.12 MB/sec
            - OverallThroughput: 5.67 MB/sec
            - PeakMemoryUsage: 456.00 B (456)
            - RowsReturned: 93 (93)
            - SerializeBatchTime: 1.026ms
            - TransmitDataRPCTime: 1.988ms
            - UncompressedRowBatchSize: 37.12 KB (38010)
         CodeGen:(Total: 68.333ms, non-child: 68.333ms, % non-child: 
100.00%)
            - CodegenTime: 1.251ms
            - CompileTime: 10.833ms
            - LoadTime: 0.000ns
            - ModuleBitcodeSize: 1.86 MB (1950884)
            - NumFunctions: 32 (32)
            - NumInstructions: 483 (483)
            - OptimizationTime: 32.844ms
            - PeakMemoryUsage: 241.50 KB (247296)
            - PrepareTime: 23.839ms
         HDFS_SCAN_NODE (id=0):(Total: 16s448ms, non-child: 16s448ms, % 
non-child: 100.00%)
           Hdfs split stats (<volume id>:<# splits>/<split lengths>): 
1:1/162.67 MB 0:1/252.38 MB 2:2/504.68 MB 4:3/756.96 MB 3:2/504.74 MB
           ExecOption: PARQUET Codegen Enabled, Codegen enabled: 9 out of 9
           Hdfs Read Thread Concurrency Bucket: 0:81.82% 1:18.18% 2:0% 
3:0% 4:0% 5:0% 6:0% 7:0% 8:0% 9:0% 10:0% 11:0%
           File Formats: PARQUET/SNAPPY:9
           BytesRead(500.000ms): 0, 134.13 MB, 186.80 MB, 234.80 MB, 
354.64 MB, 402.64 MB, 447.06 MB, 495.06 MB, 614.97 MB, 646.97 MB, 691.70 
MB, 731.70 MB, 757.29 MB, 866.45 MB, 906.45 MB, 1007.14 MB, 1.05 GB, 
1.09 GB, 1.14 GB, 1.25 GB, 1.30 GB, 1.34 GB, 1.39 GB, 1.51 GB, 1.56 GB, 
1.60 GB, 1.64 GB, 1.76 GB, 1.81 GB, 1.86 GB, 1.89 GB, 2.02 GB, 2.07 GB, 
2.11 GB
            - FooterProcessingTime: (Avg: 3.282ms ; Min: 1.994ms ; Max: 
10.196ms ; Number of samples: 9)
            - AverageHdfsReadThreadConcurrency: 0.18
            - AverageScannerThreadConcurrency: 1.00
            - BytesRead: 2.13 GB (2288275802)
            - BytesReadDataNodeCache: 0
            - BytesReadLocal: 2.13 GB (2288275802)
            - BytesReadRemoteUnexpected: 0
            - BytesReadShortCircuit: 2.13 GB (2288275802)
            - CachedFileHandlesHitCount: 0 (0)
            - CachedFileHandlesMissCount: 171 (171)
            - CollectionItemsRead: 0 (0)
            - DecompressionTime: 3s738ms
            - MaxCompressedTextFileLength: 0
            - NumColumns: 18 (18)
            - NumDictFilteredRowGroups: 0 (0)
            - NumDisksAccessed: 5 (5)
            - NumRowGroups: 9 (9)
            - NumScannerThreadsStarted: 1 (1)
            - NumScannersWithNoReads: 0 (0)
            - NumStatsFilteredRowGroups: 0 (0)
            - PeakMemoryUsage: 143.25 MB (150207517)
            - PerReadThreadRawHdfsThroughput: 1.50 GB/sec
            - RemoteScanRanges: 0 (0)
            - RowBatchQueueGetWaitTime: 16s443ms
            - RowBatchQueuePutWaitTime: 0.000ns
            - RowsRead: 51.90M (51903646)
            - RowsReturned: 93 (93)
            - RowsReturnedRate: 5.00 /sec
            - ScanRangesComplete: 9 (9)
            - ScannerThreadsInvoluntaryContextSwitches: 26 (26)
            - ScannerThreadsTotalWallClockTime: 16s884ms
              - MaterializeTupleTime(*): 12s402ms
              - ScannerThreadsSysTime: 667.204ms
              - ScannerThreadsUserTime: 15s717ms
            - ScannerThreadsVoluntaryContextSwitches: 1.05K (1054)
            - TotalRawHdfsReadTime(*): 1s422ms
            - TotalReadThroughput: 127.33 MB/sec
       Instance b4f4f6493e9fe5a:fc9961d600000001 
(host=bigdb01:22000):(Total: 15s542ms, non-child: 413.267ms, % 
non-child: 2.66%)
         Hdfs split stats (<volume id>:<# splits>/<split lengths>): 
0:1/252.36 MB 3:1/149.88 MB 2:2/504.73 MB 4:1/252.35 MB 1:4/1009.49 MB
         MemoryUsage(500.000ms): 134.10 MB, 133.28 MB, 126.63 MB, 110.79 
MB, 88.39 MB, 134.59 MB, 126.31 MB, 114.98 MB, 91.04 MB, 131.00 MB, 
127.24 MB, 120.06 MB, 90.81 MB, 97.26 MB, 83.69 MB, 129.41 MB, 128.89 
MB, 123.06 MB, 111.92 MB, 89.77 MB, 130.64 MB, 122.12 MB, 106.75 MB, 
130.91 MB, 128.53 MB, 127.94 MB, 110.48 MB, 40.90 MB, 129.82 MB, 122.63 
MB, 104.94 MB
         ThreadUsage(500.000ms): 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
          - AverageThreadTokens: 2.00
          - BloomFilterBytes: 0
          - PeakMemoryUsage: 144.09 MB (151087586)
          - PeakReservation: 0
          - PeakUsedReservation: 0
          - PerHostPeakMemUsage: 144.57 MB (151595498)
          - RowsProduced: 89 (89)
          - TotalNetworkReceiveTime: 0.000ns
          - TotalNetworkSendTime: 17.632us
          - TotalStorageWaitTime: 434.756ms
          - TotalThreadsInvoluntaryContextSwitches: 95 (95)
          - TotalThreadsTotalWallClockTime: 30s991ms
            - TotalThreadsSysTime: 1s050ms
            - TotalThreadsUserTime: 14s447ms
          - TotalThreadsVoluntaryContextSwitches: 874 (874)
         Fragment Instance Lifecycle Timings:
            - ExecTime: 15s475ms
              - ExecTreeExecTime: 15s061ms
            - OpenTime: 40.572ms
              - ExecTreeOpenTime: 41.278us
            - PrepareTime: 26.145ms
              - ExecTreePrepareTime: 222.311us
         DataStreamSender (dst_id=1):(Total: 3.198ms, non-child: 
3.198ms, % non-child: 100.00%)
            - BytesSent: 16.21 KB (16600)
            - NetworkThroughput(*): 7.50 MB/sec
            - OverallThroughput: 4.95 MB/sec
            - PeakMemoryUsage: 456.00 B (456)
            - RowsReturned: 89 (89)
            - SerializeBatchTime: 856.396us
            - TransmitDataRPCTime: 2.112ms
            - UncompressedRowBatchSize: 36.08 KB (36942)
         CodeGen:(Total: 64.650ms, non-child: 64.650ms, % non-child: 
100.00%)
            - CodegenTime: 1.228ms
            - CompileTime: 9.773ms
            - LoadTime: 0.000ns
            - ModuleBitcodeSize: 1.86 MB (1950884)
            - NumFunctions: 32 (32)
            - NumInstructions: 483 (483)
            - OptimizationTime: 29.748ms
            - PeakMemoryUsage: 241.50 KB (247296)
            - PrepareTime: 24.331ms
         HDFS_SCAN_NODE (id=0):(Total: 15s061ms, non-child: 15s061ms, % 
non-child: 100.00%)
           Hdfs split stats (<volume id>:<# splits>/<split lengths>): 
0:1/252.36 MB 3:1/149.88 MB 2:2/504.73 MB 4:1/252.35 MB 1:4/1009.49 MB
           ExecOption: PARQUET Codegen Enabled, Codegen enabled: 9 out of 9
           Hdfs Read Thread Concurrency Bucket: 0:87.1% 1:12.9% 2:0% 
3:0% 4:0% 5:0% 6:0% 7:0% 8:0% 9:0% 10:0% 11:0%
           File Formats: PARQUET/SNAPPY:8 PARQUET/(NONE,SNAPPY):1
           BytesRead(500.000ms): 102.00 MB, 150.00 MB, 194.70 MB, 226.70 
MB, 252.45 MB, 378.45 MB, 423.23 MB, 463.23 MB, 504.89 MB, 623.15 MB, 
667.91 MB, 715.91 MB, 757.35 MB, 854.18 MB, 894.18 MB, 1009.41 MB, 1.02 
GB, 1.07 GB, 1.11 GB, 1.13 GB, 1.26 GB, 1.32 GB, 1.36 GB, 1.48 GB, 1.52 
GB, 1.56 GB, 1.60 GB, 1.65 GB, 1.77 GB, 1.82 GB, 1.87 GB
            - FooterProcessingTime: (Avg: 3.554ms ; Min: 1.844ms ; Max: 
13.695ms ; Number of samples: 9)
            - AverageHdfsReadThreadConcurrency: 0.13
            - AverageScannerThreadConcurrency: 1.00
            - BytesRead: 1.88 GB (2013820022)
            - BytesReadDataNodeCache: 0
            - BytesReadLocal: 1.88 GB (2013820022)
            - BytesReadRemoteUnexpected: 0
            - BytesReadShortCircuit: 1.88 GB (2013820022)
            - CachedFileHandlesHitCount: 0 (0)
            - CachedFileHandlesMissCount: 154 (154)
            - CollectionItemsRead: 0 (0)
            - DecompressionTime: 3s506ms
            - MaxCompressedTextFileLength: 0
            - NumColumns: 18 (18)
            - NumDictFilteredRowGroups: 1 (1)
            - NumDisksAccessed: 5 (5)
            - NumRowGroups: 9 (9)
            - NumScannerThreadsStarted: 1 (1)
            - NumScannersWithNoReads: 0 (0)
            - NumStatsFilteredRowGroups: 0 (0)
            - PeakMemoryUsage: 143.15 MB (150100301)
            - PerReadThreadRawHdfsThroughput: 1.56 GB/sec
            - RemoteScanRanges: 0 (0)
            - RowBatchQueueGetWaitTime: 15s057ms
            - RowBatchQueuePutWaitTime: 0.000ns
            - RowsRead: 45.60M (45604014)
            - RowsReturned: 89 (89)
            - RowsReturnedRate: 5.00 /sec
            - ScanRangesComplete: 9 (9)
            - ScannerThreadsInvoluntaryContextSwitches: 82 (82)
            - ScannerThreadsTotalWallClockTime: 15s475ms
              - MaterializeTupleTime(*): 11s276ms
              - ScannerThreadsSysTime: 674.148ms
              - ScannerThreadsUserTime: 14s361ms
            - ScannerThreadsVoluntaryContextSwitches: 567 (567)
            - TotalRawHdfsReadTime(*): 1s202ms
            - TotalReadThroughput: 123.27 MB/sec
       Instance b4f4f6493e9fe5a:fc9961d600000003 
(host=bigdb03:22000):(Total: 13s653ms, non-child: 461.761ms, % 
non-child: 3.38%)
         Hdfs split stats (<volume id>:<# splits>/<split lengths>): 
2:2/504.66 MB 0:1/252.37 MB 3:3/654.56 MB 1:3/757.17 MB
         MemoryUsage(500.000ms): 132.79 MB, 123.57 MB, 108.74 MB, 97.90 
MB, 79.75 MB, 129.78 MB, 125.51 MB, 110.49 MB, 69.49 MB, 141.21 MB, 
114.37 MB, 89.40 MB, 130.39 MB, 121.69 MB, 92.30 MB, 135.26 MB, 127.93 
MB, 104.04 MB, 130.38 MB, 120.88 MB, 104.35 MB, 132.65 MB, 123.71 MB, 
106.08 MB, 130.69 MB, 123.63 MB, 112.19 MB
         ThreadUsage(500.000ms): 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
          - AverageThreadTokens: 2.00
          - BloomFilterBytes: 0
          - PeakMemoryUsage: 144.09 MB (151091614)
          - PeakReservation: 0
          - PeakUsedReservation: 0
          - PerHostPeakMemUsage: 144.09 MB (151091614)
          - RowsProduced: 104 (104)
          - TotalNetworkReceiveTime: 0.000ns
          - TotalNetworkSendTime: 21.748us
          - TotalStorageWaitTime: 445.522ms
          - TotalThreadsInvoluntaryContextSwitches: 20 (20)
          - TotalThreadsTotalWallClockTime: 27s208ms
            - TotalThreadsSysTime: 1s005ms
            - TotalThreadsUserTime: 12s631ms
          - TotalThreadsVoluntaryContextSwitches: 1.41K (1410)
         Fragment Instance Lifecycle Timings:
            - ExecTime: 13s582ms
              - ExecTreeExecTime: 13s119ms
            - OpenTime: 44.193ms
              - ExecTreeOpenTime: 45.276us
            - PrepareTime: 26.432ms
              - ExecTreePrepareTime: 245.688us
         DataStreamSender (dst_id=1):(Total: 3.138ms, non-child: 
3.138ms, % non-child: 100.00%)
            - BytesSent: 19.53 KB (19998)
            - NetworkThroughput(*): 11.00 MB/sec
            - OverallThroughput: 6.08 MB/sec
            - PeakMemoryUsage: 456.00 B (456)
            - RowsReturned: 104 (104)
            - SerializeBatchTime: 1.264ms
            - TransmitDataRPCTime: 1.733ms
            - UncompressedRowBatchSize: 42.10 KB (43114)
         CodeGen:(Total: 68.585ms, non-child: 68.585ms, % non-child: 
100.00%)
            - CodegenTime: 1.290ms
            - CompileTime: 10.802ms
            - LoadTime: 0.000ns
            - ModuleBitcodeSize: 1.86 MB (1950884)
            - NumFunctions: 32 (32)
            - NumInstructions: 483 (483)
            - OptimizationTime: 32.495ms
            - PeakMemoryUsage: 241.50 KB (247296)
            - PrepareTime: 24.471ms
         HDFS_SCAN_NODE (id=0):(Total: 13s119ms, non-child: 13s119ms, % 
non-child: 100.00%)
           Hdfs split stats (<volume id>:<# splits>/<split lengths>): 
2:2/504.66 MB 0:1/252.37 MB 3:3/654.56 MB 1:3/757.17 MB
           ExecOption: PARQUET Codegen Enabled, Codegen enabled: 9 out of 9
           Hdfs Read Thread Concurrency Bucket: 0:92.59% 1:7.407% 2:0% 
3:0% 4:0% 5:0% 6:0% 7:0% 8:0% 9:0% 10:0% 11:0%
           File Formats: PARQUET/SNAPPY:9
           BytesRead(500.000ms): 102.13 MB, 170.89 MB, 226.89 MB, 333.23 
MB, 389.23 MB, 512.37 MB, 565.15 MB, 621.15 MB, 696.90 MB, 804.95 MB, 
865.63 MB, 907.22 MB, 1.02 GB, 1.08 GB, 1.13 GB, 1.26 GB, 1.32 GB, 1.38 
GB, 1.50 GB, 1.56 GB, 1.62 GB, 1.74 GB, 1.80 GB, 1.85 GB, 1.99 GB, 2.04 
GB, 2.09 GB
            - FooterProcessingTime: (Avg: 4.124ms ; Min: 1.805ms ; Max: 
19.516ms ; Number of samples: 9)
            - AverageHdfsReadThreadConcurrency: 0.07
            - AverageScannerThreadConcurrency: 1.00
            - BytesRead: 2.12 GB (2275008425)
            - BytesReadDataNodeCache: 0
            - BytesReadLocal: 2.12 GB (2275008425)
            - BytesReadRemoteUnexpected: 0
            - BytesReadShortCircuit: 2.12 GB (2275008425)
            - CachedFileHandlesHitCount: 0 (0)
            - CachedFileHandlesMissCount: 171 (171)
            - CollectionItemsRead: 0 (0)
            - DecompressionTime: 3s303ms
            - MaxCompressedTextFileLength: 0
            - NumColumns: 18 (18)
            - NumDictFilteredRowGroups: 0 (0)
            - NumDisksAccessed: 4 (4)
            - NumRowGroups: 9 (9)
            - NumScannerThreadsStarted: 1 (1)
            - NumScannersWithNoReads: 0 (0)
            - NumStatsFilteredRowGroups: 0 (0)
            - PeakMemoryUsage: 143.25 MB (150212813)
            - PerReadThreadRawHdfsThroughput: 1.69 GB/sec
            - RemoteScanRanges: 0 (0)
            - RowBatchQueueGetWaitTime: 13s114ms
            - RowBatchQueuePutWaitTime: 0.000ns
            - RowsRead: 51.60M (51599256)
            - RowsReturned: 104 (104)
            - RowsReturnedRate: 7.00 /sec
            - ScanRangesComplete: 9 (9)
            - ScannerThreadsInvoluntaryContextSwitches: 17 (17)
            - ScannerThreadsTotalWallClockTime: 13s581ms
              - MaterializeTupleTime(*): 9s602ms
              - ScannerThreadsSysTime: 596.050ms
              - ScannerThreadsUserTime: 12s527ms
            - ScannerThreadsVoluntaryContextSwitches: 1.09K (1092)
            - TotalRawHdfsReadTime(*): 1s255ms
            - TotalReadThroughput: 158.81 MB/sec


On 08/03/2018 02:22 AM, Tim Armstrong wrote:
> Can you include a complete profile?
>
> On Thu, Aug 2, 2018 at 12:31 AM, meijie <meijies@yandex.com 
> <ma...@yandex.com>> wrote:
>
>     Hi, Every One
>
>     I found the sql (select * from parquet_session_1 where browser =
>     'Firefox 20') was very slow at my impala cluster. which needed 15
>     seconds to executed and NumScannerThreadsStarted is 1 , it wasn't
>     as expected.
>
>     the part profile as blew:
>
>                - FooterProcessingTime: (Avg: 4.430ms ; Min: 2.924ms ;
>     Max: 9.559ms ; Number of samples: 9)
>                - AverageHdfsReadThreadConcurrency: 0.00
>                - AverageScannerThreadConcurrency: 1.00
>                - BytesRead: 2.13 GB (2288275802)
>                - BytesReadDataNodeCache: 0
>                - BytesReadLocal: 2.13 GB (2288275802)
>                - BytesReadRemoteUnexpected: 0
>                - BytesReadShortCircuit: 2.13 GB (2288275802)
>                - CachedFileHandlesHitCount: 0 (0)
>                - CachedFileHandlesMissCount: 171 (171)
>                - CollectionItemsRead: 0 (0)
>                - DecompressionTime: 3s264ms
>                - MaxCompressedTextFileLength: 0
>                - NumColumns: 18 (18)
>                - NumDictFilteredRowGroups: 0 (0)
>                - NumDisksAccessed: 5 (5)
>                - NumRowGroups: 9 (9)
>                - NumScannerThreadsStarted: 1 (1)
>                - NumScannersWithNoReads: 0 (0)
>                - NumStatsFilteredRowGroups: 0 (0)
>                - PeakMemoryUsage: 143.25 MB (user150207517)
>                - PerReadThreadRawHdfsThroughput: 1.30 GB/sec
>                - RemoteScanRanges: 0 (0)
>                - RowBatchQueueGetWaitTime: 14s216ms
>                - RowBatchQueuePutWaitTime: 0.000ns
>                - RowsRead: 51.90M (51903646)
>                - RowsReturned: 93 (93)
>                - RowsReturnedRate: 6.00 /sec
>                - ScanRangesComplete: 9 (9)
>                - ScannerThreadsInvoluntaryContextSwitches: 18 (18)
>                - ScannerThreadsTotalWallClockTime: 14s678ms
>                  - MaterializeTupleTime(*): 10s639ms
>                  - ScannerThreadsSysTime: 698.325ms
>                  - ScannerThreadsUserTime: 13s452ms
>                - ScannerThreadsVoluntaryContextSwitches: 1.38K (1382)
>                - TotalRawHdfsReadTime(*): 1s637ms
>                - TotalReadThroughput: 148.73 MB/sec
>
>     But, if I only select one column(select uuid from
>     parquet_session_1 where browser = 'Firefox 20'), It just needed
>     0.5 seconds and NumScannerThreadsStarted is 4. I don't know what
>     affects the number of scan threads.
>
>     Impala Version: 2.11.0-cdh5.14.0 RELEASE, the sql submitted by
>     impala-shell
>
>     Thanks
>
>     -- 
>     Best Regard
>     meijie
>
>

-- 
Best Regard
meijie


Re: Slow sql because NumScannerThreadsStarted is 1

Posted by Tim Armstrong <ta...@cloudera.com>.
Can you include a complete profile?

On Thu, Aug 2, 2018 at 12:31 AM, meijie <me...@yandex.com> wrote:

> Hi, Every One
>
> I found the sql (select * from parquet_session_1 where browser = 'Firefox
> 20') was very slow at my impala cluster. which needed 15 seconds to
> executed and NumScannerThreadsStarted is 1 , it wasn't as expected.
>
> the part profile as blew:
>
>            - FooterProcessingTime: (Avg: 4.430ms ; Min: 2.924ms ; Max:
> 9.559ms ; Number of samples: 9)
>            - AverageHdfsReadThreadConcurrency: 0.00
>            - AverageScannerThreadConcurrency: 1.00
>            - BytesRead: 2.13 GB (2288275802)
>            - BytesReadDataNodeCache: 0
>            - BytesReadLocal: 2.13 GB (2288275802)
>            - BytesReadRemoteUnexpected: 0
>            - BytesReadShortCircuit: 2.13 GB (2288275802)
>            - CachedFileHandlesHitCount: 0 (0)
>            - CachedFileHandlesMissCount: 171 (171)
>            - CollectionItemsRead: 0 (0)
>            - DecompressionTime: 3s264ms
>            - MaxCompressedTextFileLength: 0
>            - NumColumns: 18 (18)
>            - NumDictFilteredRowGroups: 0 (0)
>            - NumDisksAccessed: 5 (5)
>            - NumRowGroups: 9 (9)
>            - NumScannerThreadsStarted: 1 (1)
>            - NumScannersWithNoReads: 0 (0)
>            - NumStatsFilteredRowGroups: 0 (0)
>            - PeakMemoryUsage: 143.25 MB (150207517)
>            - PerReadThreadRawHdfsThroughput: 1.30 GB/sec
>            - RemoteScanRanges: 0 (0)
>            - RowBatchQueueGetWaitTime: 14s216ms
>            - RowBatchQueuePutWaitTime: 0.000ns
>            - RowsRead: 51.90M (51903646)
>            - RowsReturned: 93 (93)
>            - RowsReturnedRate: 6.00 /sec
>            - ScanRangesComplete: 9 (9)
>            - ScannerThreadsInvoluntaryContextSwitches: 18 (18)
>            - ScannerThreadsTotalWallClockTime: 14s678ms
>              - MaterializeTupleTime(*): 10s639ms
>              - ScannerThreadsSysTime: 698.325ms
>              - ScannerThreadsUserTime: 13s452ms
>            - ScannerThreadsVoluntaryContextSwitches: 1.38K (1382)
>            - TotalRawHdfsReadTime(*): 1s637ms
>            - TotalReadThroughput: 148.73 MB/sec
>
> But, if I only select one column(select uuid from parquet_session_1 where
> browser = 'Firefox 20'), It just needed 0.5 seconds and
> NumScannerThreadsStarted is 4. I don't know what affects the number of scan
> threads.
>
> Impala Version: 2.11.0-cdh5.14.0 RELEASE, the sql submitted by impala-shell
>
> Thanks
>
> --
> Best Regard
> meijie
>
>