Skip to content

[Enhancement] remove rouitine load plan lock #50784

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

hoffermei
Copy link
Contributor

Why I'm doing:

As trace below, routine load job plan takes much time, especially in table with many tablets case. In case routine load job with many concurrent tasks, conflict on lock will be outstanding.

Affect(class count: 3 , method count: 170) cost in 1839 ms, listenerId: 26
`---ts=2024-08-16 19:58:37;thread_name=pool-17-thread-5;id=1aa54a8;is_daemon=false;priority=5;TCCL=sun.misc.Launcher$AppClassLoader@63947c6b
    `---[153.629655ms] com.starrocks.load.routineload.RoutineLoadJob:plan()
        +---[0.00% 0.001884ms ] com.starrocks.server.GlobalStateMgr:getCurrentState() #746
        +---[0.00% 0.001393ms ] com.starrocks.server.GlobalStateMgr:getDb() #746
        +---[0.00% 0.001403ms ] com.starrocks.catalog.Database:getTable() #750
        +---[0.00% 0.001974ms ] com.starrocks.load.LoadUtils:buildIcebergTableSink() #754
        +---[6.02% 9.242233ms ] com.starrocks.catalog.Database:readLock() #755
        +---[0.05% 0.072716ms ] com.starrocks.load.streamload.StreamLoadInfo:fromRoutineLoadJob() #758
        |   +---[3.42% 0.002485ms ] com.starrocks.load.routineload.RoutineLoadJob:getFormat()
        |   +---[2.12% min=6.52E-4ms,max=8.92E-4ms,total=0.001544ms,count=2] com.starrocks.load.routineload.RoutineLoadJob:getColumnDescs()
        |   +---[1.01% 7.32E-4ms ] com.starrocks.load.routineload.RoutineLoadJob:getWhereExpr()
        |   +---[1.24% 9.02E-4ms ] com.starrocks.load.routineload.RoutineLoadJob:getMergeCondition()
        |   +---[0.98% 7.11E-4ms ] com.starrocks.load.routineload.RoutineLoadJob:getColumnSeparator()
        |   +---[1.02% 7.42E-4ms ] com.starrocks.load.routineload.RoutineLoadJob:getRowDelimiter()
        |   +---[0.91% 6.61E-4ms ] com.starrocks.load.routineload.RoutineLoadJob:getPartitions()
        |   +---[3.28% 0.002385ms ] com.starrocks.load.routineload.RoutineLoadJob:isStrictMode()
        |   +---[1.61% 0.001172ms ] com.starrocks.load.routineload.RoutineLoadJob:getTimezone()
        |   +---[1.69% 0.001232ms ] com.starrocks.load.routineload.RoutineLoadJob:getJsonPaths()
        |   +---[1.39% 0.001011ms ] com.starrocks.load.routineload.RoutineLoadJob:getJsonRoot()
        |   +---[1.50% 0.001092ms ] com.starrocks.load.routineload.RoutineLoadJob:isStripOuterArray()
        |   +---[1.87% 0.001363ms ] com.starrocks.load.routineload.RoutineLoadJob:isPartialUpdate()
        |   +---[3.47% min=6.01E-4ms,max=6.81E-4ms,total=0.002524ms,count=4] com.starrocks.load.routineload.RoutineLoadJob:getSessionVariables()
        |   +---[1.79% 0.001303ms ] com.starrocks.load.routineload.RoutineLoadJob:getSinkWithBrokerName()
        |   +---[2.66% 0.001934ms ] com.starrocks.load.routineload.RoutineLoadJob:isDiscardUnknownFields()
        |   +---[1.60% 0.001162ms ] com.starrocks.load.routineload.RoutineLoadJob:isAutoFillMissingFields()
        |   `---[1.64% 0.001192ms ] com.starrocks.load.routineload.RoutineLoadJob:isEnableDeltaLHWrite()
        +---[0.01% 0.012053ms ] com.starrocks.planner.StreamLoadPlanner:<init>() #758
        +---[93.88% 144.230367ms ] com.starrocks.planner.StreamLoadPlanner:plan() #759
        +---[0.00% 0.002204ms ] com.starrocks.thrift.TQueryOptions:setLoad_job_type() #760
        +---[0.00% 0.001683ms ] com.starrocks.server.GlobalStateMgr:getCurrentGlobalTransactionMgr() #763
        +---[0.00% 0.001853ms ] com.starrocks.catalog.Database:getId() #763
        +---[0.00% 0.003156ms ] com.starrocks.transaction.GlobalTransactionMgr:getTransactionState() #763
        +---[0.00% 0.001232ms ] com.starrocks.planner.StreamLoadPlanner:getDestTable() #767
        +---[0.00% 0.004238ms ] com.starrocks.transaction.TransactionState:addTableIndexes() #769
        `---[0.01% 0.013325ms ] com.starrocks.catalog.Database:readUnlock() #773
        

`---ts=2024-08-16 20:00:56;thread_name=pool-17-thread-47;id=1aa54d7;is_daemon=false;priority=5;TCCL=sun.misc.Launcher$AppClassLoader@63947c6b
    `---[226.964576ms] com.starrocks.planner.StreamLoadPlanner:plan()
        +---[0.00% 0.001283ms ] com.starrocks.catalog.OlapTable:getKeysType() #153
        +---[0.00% 0.007694ms ] com.starrocks.planner.StreamLoadPlanner:resetAnalyzer() #155
        +---[0.00% 0.001864ms ] com.starrocks.analysis.DescriptorTable:createTupleDescriptor() #157
        +---[0.00% 0.001322ms ] com.starrocks.load.streamload.StreamLoadInfo:getNegative() #158
        +---[0.00% 0.001333ms ] com.starrocks.load.streamload.StreamLoadInfo:isPartialUpdate() #164
        +---[0.00% 0.001092ms ] com.google.common.collect.Lists:newArrayList() #168
        +---[0.00% 6.81E-4ms ] com.starrocks.load.streamload.StreamLoadInfo:isPartialUpdate() #170
        +---[0.00% 0.001282ms ] com.starrocks.catalog.Table:getFullSchema() #173
        +---[0.11% min=6.71E-4ms,max=0.018104ms,total=0.242874ms,count=265] com.starrocks.analysis.DescriptorTable:addSlotDescriptor() #176
        +---[0.09% min=5.51E-4ms,max=0.018615ms,total=0.199948ms,count=265] com.starrocks.analysis.SlotDescriptor:setIsMaterialized() #177
        +---[0.11% min=6.41E-4ms,max=0.016832ms,total=0.240397ms,count=265] com.starrocks.analysis.SlotDescriptor:setColumn() #178
        +---[0.06% min=5.21E-4ms,max=0.001242ms,total=0.144655ms,count=265] com.starrocks.catalog.Column:isAllowNull() #179
        +---[0.07% min=5.31E-4ms,max=0.021229ms,total=0.169557ms,count=265] com.starrocks.analysis.SlotDescriptor:setIsNullable() #179
        +---[0.08% min=5.51E-4ms,max=0.017352ms,total=0.178227ms,count=265] com.starrocks.catalog.Column:getType() #184
        +---[0.07% min=5.2E-4ms,max=0.008285ms,total=0.160288ms,count=265] com.starrocks.catalog.Type:isVarchar() #184
        +---[0.01% min=6.31E-4ms,max=0.001382ms,total=0.026249ms,count=39] com.starrocks.sql.optimizer.statistics.IDictManager:getInstance() #185
        +---[0.01% min=5.1E-4ms,max=0.009478ms,total=0.030535ms,count=39] com.starrocks.catalog.Table:getId() #185
        +---[0.01% min=5.71E-4ms,max=0.001122ms,total=0.024064ms,count=39] com.starrocks.catalog.Column:getName() #186
        +---[0.02% min=6.41E-4ms,max=0.015489ms,total=0.052346ms,count=39] com.starrocks.sql.optimizer.statistics.IDictManager:hasGlobalDict() #185
        +---[0.00% 0.001202ms ] com.starrocks.planner.PlanNodeId:<init>() #200
        +---[0.00% 0.003376ms ] com.starrocks.planner.StreamLoadScanNode:<init>() #200
        +---[0.00% 0.001753ms ] com.starrocks.planner.StreamLoadScanNode:setUseVectorizedLoad() #202
        +---[0.22% 0.507622ms ] com.starrocks.planner.StreamLoadScanNode:init() #203
        +---[0.11% 0.243186ms ] com.starrocks.planner.StreamLoadScanNode:finalizeStats() #204
        +---[0.01% 0.020398ms ] com.starrocks.analysis.DescriptorTable:computeMemLayout() #206
        +---[0.00% 0.001613ms ] com.starrocks.catalog.OlapTable:writeQuorum() #213
        +---[0.01% 0.029415ms ] com.starrocks.planner.StreamLoadPlanner:getAllPartitionIds() #214
        +---[0.00% 0.001593ms ] com.starrocks.catalog.OlapTable:getExternalTable() #223
        +---[0.00% 0.001322ms ] com.starrocks.load.streamload.StreamLoadInfo:isEnableDeltaLHWrite() #224
        +---[0.00% 0.001502ms ] com.starrocks.catalog.OlapTable:enableReplicatedStorage() #253
        +---[0.00% 0.001854ms ] com.starrocks.planner.OlapTableSink:<init>() #253
        +---[0.00% 0.001323ms ] com.starrocks.load.streamload.StreamLoadInfo:getTxnId() #254
        +---[0.00% 0.001132ms ] com.starrocks.catalog.Database:getId() #254
        +---[0.00% 0.001253ms ] com.starrocks.load.streamload.StreamLoadInfo:getTimeout() #254
        +---[0.00% 0.005821ms ] com.starrocks.planner.OlapTableSink:init() #254
        +---[0.00% 0.001272ms ] com.starrocks.load.streamload.StreamLoadInfo:getMergeConditionStr() #255
        +---[0.00% 0.001322ms ] com.starrocks.load.Load:checkMergeCondition() #255
        +---[0.00% 8.51E-4ms ] com.starrocks.load.streamload.StreamLoadInfo:getMergeConditionStr() #256
        +---[98.15% 222.766132ms ] com.starrocks.planner.OlapTableSink:complete() #256
        +---[0.00% 0.001883ms ] com.starrocks.planner.PlanFragmentId:<init>() #282
        +---[0.00% 0.003367ms ] com.starrocks.planner.PlanFragment:<init>() #282
        +---[0.00% 0.002054ms ] com.starrocks.planner.PlanFragment:setSink() #283
        +---[0.00% 0.001042ms ] com.starrocks.planner.PlanFragment:setPipelineDop() #288
        +---[0.00% 7.52E-4ms ] com.starrocks.planner.PlanFragment:setSink() #289
        +---[0.00% 0.001613ms ] com.starrocks.planner.PlanFragment:setLoadGlobalDicts() #297
        +---[0.00% 0.001392ms ] com.starrocks.planner.PlanFragment:createDataSink() #299
        +---[0.00% 0.001322ms ] com.starrocks.thrift.TExecPlanFragmentParams:<init>() #301
        
        
`---ts=2024-08-28 15:14:31;thread_name=pool-17-thread-1;id=4dd7b8;is_daemon=false;priority=5;TCCL=sun.misc.Launcher$AppClassLoader@63947c6b
    `---[314.711223ms] com.starrocks.planner.OlapTableSink:createLocation()
        `---[100.00% 314.701295ms ] com.starrocks.planner.OlapTableSink:createLocation() #391
            `---[100.00% 314.689062ms ] com.starrocks.planner.DataSink:createLocation()
                +---[0.00% 8.31E-4ms ] com.starrocks.thrift.TOlapTableLocationParam:<init>() #117
                +---[0.00% 9.21E-4ms ] com.google.common.collect.HashMultimap:create() #119
                +---[0.00% 6.11E-4ms ] com.starrocks.server.GlobalStateMgr:getCurrentState() #121
                +---[0.00% 0.001683ms ] com.starrocks.server.GlobalStateMgr:getOrCreateSystemInfo() #122
                +---[0.01% min=7.62E-4ms,max=0.002765ms,total=0.025407ms,count=11] com.starrocks.catalog.OlapTable:getPartition() #124
                +---[0.00% min=4.11E-4ms,max=7.31E-4ms,total=0.00511ms,count=11] com.starrocks.catalog.OlapTable:getPartitionInfo() #125
                +---[0.00% min=3.2E-4ms,max=5.41E-4ms,total=0.004088ms,count=11] com.starrocks.catalog.Partition:getId() #125
                +---[0.00% min=4.11E-4ms,max=6.91E-4ms,total=0.005551ms,count=11] com.starrocks.catalog.OlapTable:writeQuorum() #125
                +---[0.00% min=7.11E-4ms,max=0.001092ms,total=0.009489ms,count=11] com.starrocks.catalog.PartitionInfo:getQuorumNum() #125
                +---[0.01% min=9.12E-4ms,max=0.002354ms,total=0.0177ms,count=11] com.starrocks.catalog.Partition:getMaterializedIndices() #126
                +---[0.00% min=3.6E-4ms,max=6.52E-4ms,total=0.004668ms,count=11] com.starrocks.catalog.MaterializedIndex:getTablets() #127
                +---[2.14% min=2.8E-4ms,max=0.045635ms,total=6.740494ms,count=22528] com.starrocks.catalog.OlapTable:isLakeTable() #128
                +---[2.19% min=2.8E-4ms,max=0.011712ms,total=6.882584ms,count=22528] com.starrocks.catalog.OlapTable:getClusterId() #136
                +---[14.36% min=8.21E-4ms,max=0.581831ms,total=45.190484ms,count=22528] com.starrocks.catalog.LocalTablet:getNormalReplicaBackendPathMap() #136
                +---[2.33% min=3.0E-4ms,max=0.009458ms,total=7.339949ms,count=22528] com.google.common.collect.Multimap:keySet() #137
                +---[2.32% min=3.0E-4ms,max=0.01069ms,total=7.302429ms,count=22528] com.google.common.collect.Multimap:keySet() #144
                +---[3.71% min=4.7E-4ms,max=0.01608ms,total=11.668303ms,count=22528] com.google.common.collect.Lists:newArrayList() #144
                +---[2.09% min=2.7E-4ms,max=0.014968ms,total=6.578761ms,count=22528] com.starrocks.catalog.Tablet:getId() #176
                +---[2.37% min=3.0E-4ms,max=0.009588ms,total=7.444809ms,count=22528] com.starrocks.thrift.TTabletLocation:<init>() #177
                +---[2.84% min=3.7E-4ms,max=0.027272ms,total=8.947718ms,count=22528] com.starrocks.thrift.TOlapTableLocationParam:addToTablets() #175
                +---[2.42% min=3.2E-4ms,max=0.013275ms,total=7.630027ms,count=22528] com.google.common.collect.Multimap:entries() #178
                +---[6.51% min=2.8E-4ms,max=0.018976ms,total=20.48902ms,count=67584] com.starrocks.catalog.Replica:getBackendId() #179
                +---[9.44% min=3.6E-4ms,max=0.019376ms,total=29.716088ms,count=67584] com.google.common.collect.Multimap:put() #179
                +---[0.00% 7.91E-4ms ] com.starrocks.server.GlobalStateMgr:getCurrentSystemInfo() #188
                +---[0.02% 0.063088ms ] com.starrocks.system.SystemInfoService:checkExceedDiskCapacityLimit() #188
                `---[0.00% 8.71E-4ms ] com.starrocks.common.Status:ok() #189

What I'm doing:

move routine load plan out of read lock (table lock in 3.3 and db lock in 2.5), by only make a shadow copy of table in read lock, then generate plan by use table's shadow copy as parameter without read lock.

What type of PR is this:

  • BugFix
  • Feature
  • Enhancement
  • Refactor
  • UT
  • Doc
  • Tool

Does this PR entail a change in behavior?

  • Yes, this PR will result in a change in behavior.
  • No, this PR will not result in a change in behavior.

If yes, please specify the type of change:

  • Interface/UI changes: syntax, type conversion, expression evaluation, display information
  • Parameter changes: default values, similar parameters but with different default values
  • Policy changes: use new policy to replace old one, functionality automatically enabled
  • Feature removed
  • Miscellaneous: upgrade & downgrade compatibility, etc.

Checklist:

  • I have added test cases for my bug fix or my new feature
  • This pr needs user documentation (for new or modified features or behaviors)
    • I have added documentation for my new feature or new function
  • This is a backport pr

Bugfix cherry-pick branch check:

  • I have checked the version labels which the pr will be auto-backported to the target branch
    • 3.3
    • 3.2
    • 3.1
    • 3.0
    • 2.5

@hoffermei hoffermei force-pushed the remove_routine_load_plan_lock branch from 13e3acd to 66908f3 Compare September 6, 2024 03:48
Copy link

sonarqubecloud bot commented Sep 6, 2024

Copy link

github-actions bot commented Sep 6, 2024

[Java-Extensions Incremental Coverage Report]

pass : 0 / 0 (0%)

Copy link

github-actions bot commented Sep 6, 2024

[FE Incremental Coverage Report]

pass : 30 / 35 (85.71%)

file detail

path covered_line new_line coverage not_covered_line_detail
🔵 com/starrocks/load/routineload/RoutineLoadJob.java 24 29 82.76% [910, 912, 913, 914, 916]
🔵 com/starrocks/catalog/OlapTable.java 6 6 100.00% []

Copy link

github-actions bot commented Sep 6, 2024

[BE Incremental Coverage Report]

pass : 0 / 0 (0%)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant