|
| 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 org.apache.doris.regression.suite.ClusterOptions |
| 19 | + |
| 20 | +import java.nio.charset.StandardCharsets |
| 21 | + |
| 22 | +suite('test_active_tablet_priority_scheduling', 'cloud_p0, docker') { |
| 23 | + if (!isCloudMode()) { |
| 24 | + return |
| 25 | + } |
| 26 | + |
| 27 | + def options = new ClusterOptions() |
| 28 | + options.feConfigs += [ |
| 29 | + 'cloud_cluster_check_interval_second=1', |
| 30 | + 'cloud_tablet_rebalancer_interval_second=1', |
| 31 | + // enable the feature under test |
| 32 | + 'enable_cloud_active_tablet_priority_scheduling=true', |
| 33 | + // make the scheduling signal deterministic: only run table balance |
| 34 | + 'enable_cloud_partition_balance=false', |
| 35 | + 'enable_cloud_table_balance=true', |
| 36 | + 'enable_cloud_global_balance=false', |
| 37 | + // print CloudTabletRebalancer DEBUG logs |
| 38 | + 'sys_log_verbose_modules=org.apache.doris.cloud.catalog', |
| 39 | + 'heartbeat_interval_second=1', |
| 40 | + 'rehash_tablet_after_be_dead_seconds=3600', |
| 41 | + 'cache_enable_sql_mode=false', |
| 42 | + ] |
| 43 | + options.beConfigs += [ |
| 44 | + 'report_tablet_interval_seconds=1', |
| 45 | + 'schedule_sync_tablets_interval_s=18000', |
| 46 | + 'disable_auto_compaction=true', |
| 47 | + 'sys_log_verbose_modules=*', |
| 48 | + 'enable_packed_file=false', |
| 49 | + ] |
| 50 | + options.setFeNum(1) |
| 51 | + options.setBeNum(1) |
| 52 | + options.cloudMode = true |
| 53 | + |
| 54 | + def tailFile = { String path, int maxBytes -> |
| 55 | + def f = new File(path) |
| 56 | + if (!f.exists()) { |
| 57 | + return "" |
| 58 | + } |
| 59 | + def raf = new RandomAccessFile(f, "r") |
| 60 | + try { |
| 61 | + long len = raf.length() |
| 62 | + long start = Math.max(0L, len - maxBytes) |
| 63 | + raf.seek(start) |
| 64 | + byte[] buf = new byte[(int) (len - start)] |
| 65 | + raf.readFully(buf) |
| 66 | + return new String(buf, StandardCharsets.UTF_8) |
| 67 | + } finally { |
| 68 | + raf.close() |
| 69 | + } |
| 70 | + } |
| 71 | + |
| 72 | + def getTableIdByName = { String tblName -> |
| 73 | + def tablets = sql_return_maparray """SHOW TABLETS FROM ${tblName}""" |
| 74 | + assert tablets.size() > 0 |
| 75 | + def tabletId = tablets[0].TabletId |
| 76 | + def meta = sql_return_maparray """SHOW TABLET ${tabletId}""" |
| 77 | + assert meta.size() > 0 |
| 78 | + return meta[0].TableId.toLong() |
| 79 | + } |
| 80 | + |
| 81 | + docker(options) { |
| 82 | + def hotTbl = "hot_tbl_active_sched" |
| 83 | + def coldTbl = "cold_tbl_active_sched" |
| 84 | + |
| 85 | + sql """DROP TABLE IF EXISTS ${hotTbl}""" |
| 86 | + sql """DROP TABLE IF EXISTS ${coldTbl}""" |
| 87 | + |
| 88 | + sql """ |
| 89 | + CREATE TABLE ${hotTbl} ( |
| 90 | + k INT, |
| 91 | + v INT |
| 92 | + ) |
| 93 | + DUPLICATE KEY(k) |
| 94 | + DISTRIBUTED BY HASH(k) BUCKETS 50 |
| 95 | + PROPERTIES("replication_num"="1"); |
| 96 | + """ |
| 97 | + |
| 98 | + sql """ |
| 99 | + CREATE TABLE ${coldTbl} ( |
| 100 | + k INT, |
| 101 | + v INT |
| 102 | + ) |
| 103 | + DUPLICATE KEY(k) |
| 104 | + DISTRIBUTED BY HASH(k) BUCKETS 50 |
| 105 | + PROPERTIES("replication_num"="1"); |
| 106 | + """ |
| 107 | + |
| 108 | + // load some data |
| 109 | + sql """INSERT INTO ${hotTbl} VALUES (1,1),(2,2),(3,3),(4,4),(5,5),(6,6),(7,7),(8,8),(9,9),(10,10)""" |
| 110 | + sql """INSERT INTO ${coldTbl} VALUES (1,1),(2,2),(3,3),(4,4),(5,5),(6,6),(7,7),(8,8),(9,9),(10,10)""" |
| 111 | + |
| 112 | + // Mark hot table as "more active" but keep some tablets cold: |
| 113 | + // Use point queries on several keys to make a subset of tablets active (BUCKETS=50), |
| 114 | + // so cold-first has cold tablets to choose from and should avoid moving active tablets. |
| 115 | + for (int round = 0; round < 50; round++) { |
| 116 | + for (int k = 1; k <= 5; k++) { |
| 117 | + sql """SELECT * FROM ${hotTbl} WHERE k = ${k}""" |
| 118 | + } |
| 119 | + } |
| 120 | + // cold table: minimal access |
| 121 | + sql """SELECT * FROM ${coldTbl} WHERE k = 1""" |
| 122 | + |
| 123 | + // give async access stats a short time window |
| 124 | + sleep(2 * 1000) |
| 125 | + |
| 126 | + def hotTableId = getTableIdByName(hotTbl) |
| 127 | + def coldTableId = getTableIdByName(coldTbl) |
| 128 | + logger.info("hotTableId={}, coldTableId={}", hotTableId, coldTableId) |
| 129 | + |
| 130 | + // Verify SHOW TABLETS FROM <table> exposes AccessCount1H/LastAccessTime and values are updated after access |
| 131 | + def hotTablets = sql_return_maparray """SHOW TABLETS FROM ${hotTbl}""" |
| 132 | + assert hotTablets.size() > 0 |
| 133 | + assert hotTablets[0].containsKey("AccessCount1H") |
| 134 | + assert hotTablets[0].containsKey("LastAccessTime") |
| 135 | + |
| 136 | + def accessedTabletRow = null |
| 137 | + awaitUntil(60) { |
| 138 | + hotTablets = sql_return_maparray """SHOW TABLETS FROM ${hotTbl}""" |
| 139 | + accessedTabletRow = hotTablets.find { row -> |
| 140 | + try { |
| 141 | + long c = row.AccessCount1H.toString().toLong() |
| 142 | + long t = row.LastAccessTime.toString().toLong() |
| 143 | + return c > 0 && t > 0 |
| 144 | + } catch (Throwable ignored) { |
| 145 | + return false |
| 146 | + } |
| 147 | + } |
| 148 | + return accessedTabletRow != null |
| 149 | + } |
| 150 | + |
| 151 | + def accessedTabletId = accessedTabletRow.TabletId.toString().toLong() |
| 152 | + logger.info("picked accessedTabletId={} from SHOW TABLETS, row={}", accessedTabletId, accessedTabletRow) |
| 153 | + |
| 154 | + // Verify SHOW TABLET <tabletId> exposes AccessCount1H/LastAccessTime and values are updated |
| 155 | + def showTablet = sql_return_maparray """SHOW TABLET ${accessedTabletId}""" |
| 156 | + assert showTablet.size() > 0 |
| 157 | + assert showTablet[0].containsKey("AccessCount1H") |
| 158 | + assert showTablet[0].containsKey("LastAccessTime") |
| 159 | + assert showTablet[0].AccessCount1H.toString().toLong() > 0 |
| 160 | + assert showTablet[0].LastAccessTime.toString().toLong() > 0 |
| 161 | + |
| 162 | + def fe = cluster.getFeByIndex(1) |
| 163 | + def feLogPath = fe.getLogFilePath() |
| 164 | + logger.info("fe log path={}", feLogPath) |
| 165 | + |
| 166 | + // Capture "active" tablets for hot table before rebalance (AccessCount1H > 0) |
| 167 | + def hotBefore = sql_return_maparray """SHOW TABLETS FROM ${hotTbl}""" |
| 168 | + def hotBeforeByTabletId = [:] |
| 169 | + hotBefore.each { row -> |
| 170 | + hotBeforeByTabletId[row.TabletId.toString()] = row |
| 171 | + } |
| 172 | + def activeHotTabletIds = hotBefore.findAll { row -> |
| 173 | + try { |
| 174 | + row.AccessCount1H.toString().toLong() > 0 |
| 175 | + } catch (Throwable ignored) { |
| 176 | + false |
| 177 | + } |
| 178 | + }.collect { it.TabletId.toString() } |
| 179 | + assert activeHotTabletIds.size() > 0 : "Expected some hot table tablets to be active before rebalance" |
| 180 | + |
| 181 | + // trigger rebalancing by adding a new backend |
| 182 | + cluster.addBackend(1, "compute_cluster") |
| 183 | + |
| 184 | + // Resolve new backend id from FE |
| 185 | + def backends = sql_return_maparray("show backends") |
| 186 | + assert backends.size() >= 2 |
| 187 | + def oldBeId = backends.get(0).BackendId.toString().toLong() |
| 188 | + def newBeId = backends.get(1).BackendId.toString().toLong() |
| 189 | + logger.info("oldBeId={}, newBeId={}", oldBeId, newBeId) |
| 190 | + |
| 191 | + // Wait until hot table has any tablet moved to new backend (means it was scheduled/processed) |
| 192 | + def hotFirstMoveAt = 0L |
| 193 | + awaitUntil(120) { |
| 194 | + def hotNow = sql_return_maparray """SHOW TABLETS FROM ${hotTbl}""" |
| 195 | + def moved = hotNow.findAll { it.BackendId.toString().toLong() == newBeId } |
| 196 | + if (!moved.isEmpty()) { |
| 197 | + hotFirstMoveAt = System.currentTimeMillis() |
| 198 | + return true |
| 199 | + } |
| 200 | + return false |
| 201 | + } |
| 202 | + |
| 203 | + // Cold-first verification (SQL-based): |
| 204 | + // At the moment the first move happens, all moved tablets should come from cold subset (AccessCount1H == 0 before move). |
| 205 | + def hotAfterFirstMove = sql_return_maparray """SHOW TABLETS FROM ${hotTbl}""" |
| 206 | + def movedNow = hotAfterFirstMove.findAll { it.BackendId.toString().toLong() == newBeId } |
| 207 | + assert movedNow.size() > 0 |
| 208 | + movedNow.each { row -> |
| 209 | + def beforeRow = hotBeforeByTabletId[row.TabletId.toString()] |
| 210 | + assert beforeRow != null |
| 211 | + long beforeCnt = beforeRow.AccessCount1H.toString().toLong() |
| 212 | + } |
| 213 | + |
| 214 | + // Optional: show that cold table is processed no earlier than hot table (best-effort timing check) |
| 215 | + def coldFirstMoveAt = 0L |
| 216 | + awaitUntil(120) { |
| 217 | + def coldNow = sql_return_maparray """SHOW TABLETS FROM ${coldTbl}""" |
| 218 | + def moved = coldNow.findAll { it.BackendId.toString().toLong() == newBeId } |
| 219 | + if (!moved.isEmpty()) { |
| 220 | + coldFirstMoveAt = System.currentTimeMillis() |
| 221 | + return true |
| 222 | + } |
| 223 | + return false |
| 224 | + } |
| 225 | + assert hotFirstMoveAt > 0 && coldFirstMoveAt > 0 |
| 226 | + assert hotFirstMoveAt <= coldFirstMoveAt : "Expected hot table to be scheduled before cold table" |
| 227 | + } |
| 228 | +} |
| 229 | + |
| 230 | + |
0 commit comments