|
| 1 | +// Licensed to the Apache Software Foundation (ASF) under one |
| 2 | +// or more contributor license agreements. See the NOTICE file |
| 3 | +// distributed with this work for additional information |
| 4 | +// regarding copyright ownership. The ASF licenses this file |
| 5 | +// to you under the Apache License, Version 2.0 (the |
| 6 | +// "License"); you may not use this file except in compliance |
| 7 | +// with the License. You may obtain a copy of the License at |
| 8 | +// |
| 9 | +// http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | +// |
| 11 | +// Unless required by applicable law or agreed to in writing, |
| 12 | +// software distributed under the License is distributed on an |
| 13 | +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
| 14 | +// KIND, either express or implied. See the License for the |
| 15 | +// specific language governing permissions and limitations |
| 16 | +// under the License. |
| 17 | + |
| 18 | +import groovy.json.JsonSlurper |
| 19 | +import org.apache.doris.regression.suite.ClusterOptions |
| 20 | +import org.apache.doris.regression.util.DebugPoint |
| 21 | +import org.apache.doris.regression.util.NodeType |
| 22 | + |
| 23 | +suite('test_cloud_cumu_compaction_global_lock_thread_count', 'docker') { |
| 24 | + def options = new ClusterOptions() |
| 25 | + options.cloudMode = true |
| 26 | + options.enableDebugPoints() |
| 27 | + options.beConfigs += [ |
| 28 | + 'enable_java_support=false', |
| 29 | + 'cumulative_compaction_min_deltas=2', |
| 30 | + 'cumulative_compaction_max_deltas=3', |
| 31 | + 'max_cumu_compaction_threads=2', |
| 32 | + 'large_cumu_compaction_task_min_thread_num=2', |
| 33 | + 'large_cumu_compaction_task_row_num_threshold=1', |
| 34 | + 'disable_auto_compaction=true', |
| 35 | + ] |
| 36 | + options.beNum = 3 |
| 37 | + |
| 38 | + docker(options) { |
| 39 | + GetDebugPoint().clearDebugPointsForAllBEs() |
| 40 | + def backends = sql_return_maparray('SHOW BACKENDS') |
| 41 | + def lockHolderBe = backends[0] |
| 42 | + def lockFailedBe = backends[1] |
| 43 | + assertNotNull(lockHolderBe) |
| 44 | + assertNotNull(lockFailedBe) |
| 45 | + |
| 46 | + def getCompactionStatus = { be, tabletId -> |
| 47 | + def (code, out, err) = be_get_compaction_status(be.Host, be.HttpPort, tabletId) |
| 48 | + assertEquals(0, code, "failed to get compaction status: ${out}, ${err}") |
| 49 | + return new JsonSlurper().parseText(out.trim()) |
| 50 | + } |
| 51 | + |
| 52 | + def waitForCompactionRunning = { be, tabletId -> |
| 53 | + def deadline = System.currentTimeMillis() + 30000 |
| 54 | + while (System.currentTimeMillis() < deadline) { |
| 55 | + if (getCompactionStatus(be, tabletId).run_status) { |
| 56 | + return |
| 57 | + } |
| 58 | + sleep(100) |
| 59 | + } |
| 60 | + assertTrue(false, "compaction did not start for tablet ${tabletId}") |
| 61 | + } |
| 62 | + |
| 63 | + def waitForCompactionFinished = { be, tabletId -> |
| 64 | + def deadline = System.currentTimeMillis() + 30000 |
| 65 | + while (System.currentTimeMillis() < deadline) { |
| 66 | + if (!getCompactionStatus(be, tabletId).run_status) { |
| 67 | + return |
| 68 | + } |
| 69 | + sleep(100) |
| 70 | + } |
| 71 | + assertTrue(false, "compaction did not finish for tablet ${tabletId}") |
| 72 | + } |
| 73 | + |
| 74 | + def getTabletStatus = { be, tabletId -> |
| 75 | + def (code, out, err) = be_show_tablet_status(be.Host, be.HttpPort, tabletId) |
| 76 | + assertEquals(0, code, "failed to get tablet status: ${out}, ${err}") |
| 77 | + return new JsonSlurper().parseText(out.trim()) |
| 78 | + } |
| 79 | + |
| 80 | + def getRowsetCount = { be, tabletId -> |
| 81 | + def status = getTabletStatus(be, tabletId) |
| 82 | + assertTrue(status.rowsets instanceof List) |
| 83 | + return status.rowsets.size() |
| 84 | + } |
| 85 | + |
| 86 | + def createTable = { table -> |
| 87 | + sql "DROP TABLE IF EXISTS ${table} FORCE" |
| 88 | + sql """ |
| 89 | + CREATE TABLE ${table} ( |
| 90 | + k INT, |
| 91 | + v INT |
| 92 | + ) DUPLICATE KEY(k) |
| 93 | + DISTRIBUTED BY HASH(k) BUCKETS 1 |
| 94 | + PROPERTIES ( |
| 95 | + "replication_num" = "1", |
| 96 | + "disable_auto_compaction" = "true" |
| 97 | + ) |
| 98 | + """ |
| 99 | + for (int value = 0; value < 5; value++) { |
| 100 | + sql "INSERT INTO ${table} VALUES (${value}, ${value})" |
| 101 | + } |
| 102 | + sql "SELECT COUNT(*) FROM ${table}" |
| 103 | + def tablet = sql_return_maparray("SHOW TABLETS FROM ${table}")[0] |
| 104 | + return tablet.TabletId |
| 105 | + } |
| 106 | + |
| 107 | + def runCumulativeCompaction = { be, tabletId -> |
| 108 | + def (code, out, err) = be_run_cumulative_compaction(be.Host, be.HttpPort, tabletId) |
| 109 | + assertEquals(0, code, "failed to submit cumulative compaction: ${out}, ${err}") |
| 110 | + } |
| 111 | + |
| 112 | + def conflictTabletId = createTable('test_cloud_cumu_compaction_global_lock_conflict') |
| 113 | + def conflictRowsetsBefore = getRowsetCount(lockFailedBe, conflictTabletId) |
| 114 | + def holderRowsetsBefore = getRowsetCount(lockHolderBe, conflictTabletId) |
| 115 | + def blockModifyRowsets = 'CloudCumulativeCompaction::modify_rowsets.enable_spin_wait' |
| 116 | + def blockModifyRowsetsSwitch = 'CloudCumulativeCompaction::modify_rowsets.block' |
| 117 | + def holdTaskAfterExecution = 'CloudStorageEngine._submit_cumulative_compaction_task.sleep' |
| 118 | + |
| 119 | + try { |
| 120 | + DebugPoint.enableDebugPoint(lockHolderBe.Host, lockHolderBe.HttpPort.toInteger(), |
| 121 | + NodeType.BE, blockModifyRowsets) |
| 122 | + DebugPoint.enableDebugPoint(lockHolderBe.Host, lockHolderBe.HttpPort.toInteger(), |
| 123 | + NodeType.BE, blockModifyRowsetsSwitch) |
| 124 | + |
| 125 | + runCumulativeCompaction(lockHolderBe, conflictTabletId) |
| 126 | + waitForCompactionRunning(lockHolderBe, conflictTabletId) |
| 127 | + |
| 128 | + runCumulativeCompaction(lockFailedBe, conflictTabletId) |
| 129 | + sleep(1000) |
| 130 | + waitForCompactionFinished(lockFailedBe, conflictTabletId) |
| 131 | + assertEquals(conflictRowsetsBefore, getRowsetCount(lockFailedBe, conflictTabletId)) |
| 132 | + |
| 133 | + DebugPoint.disableDebugPoint(lockHolderBe.Host, lockHolderBe.HttpPort.toInteger(), |
| 134 | + NodeType.BE, blockModifyRowsetsSwitch) |
| 135 | + DebugPoint.disableDebugPoint(lockHolderBe.Host, lockHolderBe.HttpPort.toInteger(), |
| 136 | + NodeType.BE, blockModifyRowsets) |
| 137 | + waitForCompactionFinished(lockHolderBe, conflictTabletId) |
| 138 | + assertTrue(getRowsetCount(lockHolderBe, conflictTabletId) < holderRowsetsBefore) |
| 139 | + |
| 140 | + def runningTabletId = createTable('test_cloud_cumu_compaction_global_lock_thread_holder') |
| 141 | + def candidateTabletId = createTable('test_cloud_cumu_compaction_global_lock_thread_candidate') |
| 142 | + def candidateRowsetsBefore = getRowsetCount(lockFailedBe, candidateTabletId) |
| 143 | + def runningRowsetsBefore = getRowsetCount(lockFailedBe, runningTabletId) |
| 144 | + |
| 145 | + DebugPoint.enableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 146 | + NodeType.BE, holdTaskAfterExecution) |
| 147 | + DebugPoint.enableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 148 | + NodeType.BE, blockModifyRowsets) |
| 149 | + DebugPoint.enableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 150 | + NodeType.BE, blockModifyRowsetsSwitch) |
| 151 | + |
| 152 | + runCumulativeCompaction(lockFailedBe, runningTabletId) |
| 153 | + waitForCompactionRunning(lockFailedBe, runningTabletId) |
| 154 | + |
| 155 | + DebugPoint.disableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 156 | + NodeType.BE, blockModifyRowsetsSwitch) |
| 157 | + DebugPoint.disableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 158 | + NodeType.BE, blockModifyRowsets) |
| 159 | + def deadline = System.currentTimeMillis() + 30000 |
| 160 | + while (System.currentTimeMillis() < deadline && |
| 161 | + getRowsetCount(lockFailedBe, runningTabletId) >= runningRowsetsBefore) { |
| 162 | + sleep(100) |
| 163 | + } |
| 164 | + assertTrue(getRowsetCount(lockFailedBe, runningTabletId) < runningRowsetsBefore) |
| 165 | + assertTrue(getCompactionStatus(lockFailedBe, runningTabletId).run_status) |
| 166 | + |
| 167 | + runCumulativeCompaction(lockFailedBe, candidateTabletId) |
| 168 | + sleep(1000) |
| 169 | + assertEquals(candidateRowsetsBefore, getRowsetCount(lockFailedBe, candidateTabletId), |
| 170 | + 'a second large compaction must be delayed while the first task is active') |
| 171 | + waitForCompactionFinished(lockFailedBe, candidateTabletId) |
| 172 | + waitForCompactionFinished(lockFailedBe, runningTabletId) |
| 173 | + } finally { |
| 174 | + DebugPoint.disableDebugPoint(lockHolderBe.Host, lockHolderBe.HttpPort.toInteger(), |
| 175 | + NodeType.BE, blockModifyRowsetsSwitch) |
| 176 | + DebugPoint.disableDebugPoint(lockHolderBe.Host, lockHolderBe.HttpPort.toInteger(), |
| 177 | + NodeType.BE, blockModifyRowsets) |
| 178 | + DebugPoint.disableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 179 | + NodeType.BE, holdTaskAfterExecution) |
| 180 | + DebugPoint.disableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 181 | + NodeType.BE, blockModifyRowsetsSwitch) |
| 182 | + DebugPoint.disableDebugPoint(lockFailedBe.Host, lockFailedBe.HttpPort.toInteger(), |
| 183 | + NodeType.BE, blockModifyRowsets) |
| 184 | + } |
| 185 | + } |
| 186 | +} |
0 commit comments