[
https://issues.apache.org/jira/browse/HIVE-27020?focusedWorklogId=857257&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-857257
]
ASF GitHub Bot logged work on HIVE-27020:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 16/Apr/23 18:28
Start Date: 16/Apr/23 18:28
Worklog Time Spent: 10m
Work Description: SourabhBadhya commented on code in PR #4091:
URL: https://github.com/apache/hive/pull/4091#discussion_r1167994197
##########
ql/src/java/org/apache/hadoop/hive/ql/txn/compactor/handler/AbortedTxnCleaner.java:
##########
@@ -0,0 +1,168 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.hadoop.hive.ql.txn.compactor.handler;
+
+import org.apache.hadoop.hive.common.ValidReaderWriteIdList;
+import org.apache.hadoop.hive.common.ValidTxnList;
+import org.apache.hadoop.hive.conf.HiveConf;
+import org.apache.hadoop.hive.metastore.api.MetaException;
+import org.apache.hadoop.hive.metastore.api.Partition;
+import org.apache.hadoop.hive.metastore.api.Table;
+import org.apache.hadoop.hive.metastore.metrics.MetricsConstants;
+import org.apache.hadoop.hive.metastore.metrics.PerfLogger;
+import org.apache.hadoop.hive.metastore.txn.AcidTxnInfo;
+import org.apache.hadoop.hive.metastore.txn.TxnStore;
+import org.apache.hadoop.hive.metastore.txn.TxnUtils;
+import org.apache.hadoop.hive.metastore.utils.MetaStoreUtils;
+import org.apache.hadoop.hive.ql.txn.compactor.CompactorUtil;
+import org.apache.hadoop.hive.ql.txn.compactor.CompactorUtil.ThrowingRunnable;
+import org.apache.hadoop.hive.ql.txn.compactor.FSRemover;
+import org.apache.hadoop.hive.ql.txn.compactor.MetadataCache;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.util.Collections;
+import java.util.List;
+import java.util.concurrent.TimeUnit;
+import java.util.stream.Collectors;
+
+import static java.util.Objects.isNull;
+
+/**
+ * Abort-cleanup based implementation of TaskHandler.
+ * Provides implementation of creation of abort clean tasks.
+ */
+class AbortedTxnCleaner extends AcidTxnCleaner {
+
+ private static final Logger LOG =
LoggerFactory.getLogger(AbortedTxnCleaner.class.getName());
+
+ public AbortedTxnCleaner(HiveConf conf, TxnStore txnHandler,
+ MetadataCache metadataCache, boolean metricsEnabled,
+ FSRemover fsRemover) {
+ super(conf, txnHandler, metadataCache, metricsEnabled, fsRemover);
+ }
+
+ /**
+ The following cleanup is based on the following idea - <br>
+ 1. Aborted cleanup is independent of compaction. This is because
directories which are written by
+ aborted txns are not visible by any open txns. It is only visible while
determining the AcidState (which
+ only sees the aborted deltas and does not read the file).<br><br>
+
+ The following algorithm is used to clean the set of aborted directories -
<br>
+ a. Find the list of entries which are suitable for cleanup (This is done
in {@link TxnStore#findReadyToCleanForAborts(long, int)}).<br>
+ b. If the table/partition does not exist, then remove the associated
aborted entry in TXN_COMPONENTS table. <br>
+ c. Get the AcidState of the table by using the min open txnID, database
name, tableName, partition name, highest write ID <br>
+ d. Fetch the aborted directories and delete the directories. <br>
+ e. Fetch the aborted write IDs from the AcidState and use it to delete
the associated metadata in the TXN_COMPONENTS table.
+ **/
+ @Override
+ public List<Runnable> getTasks() throws MetaException {
+ int abortedThreshold = HiveConf.getIntVar(conf,
+ HiveConf.ConfVars.HIVE_COMPACTOR_ABORTEDTXN_THRESHOLD);
+ long abortedTimeThreshold = HiveConf
+ .getTimeVar(conf,
HiveConf.ConfVars.HIVE_COMPACTOR_ABORTEDTXN_TIME_THRESHOLD,
+ TimeUnit.MILLISECONDS);
+ List<AcidTxnInfo> readyToCleanAborts =
txnHandler.findReadyToCleanForAborts(abortedTimeThreshold, abortedThreshold);
+
+ if (!readyToCleanAborts.isEmpty()) {
+ return readyToCleanAborts.stream().map(ci ->
ThrowingRunnable.unchecked(() ->
+ clean(ci, ci.txnId > 0 ? ci.txnId : Long.MAX_VALUE,
metricsEnabled)))
+ .collect(Collectors.toList());
+ }
+ return Collections.emptyList();
+ }
+
+ private void clean(AcidTxnInfo info, long minOpenTxn, boolean
metricsEnabled) throws MetaException {
+ LOG.info("Starting cleaning for {}", info);
+ PerfLogger perfLogger = PerfLogger.getPerfLogger(false);
+ String cleanerMetric = MetricsConstants.COMPACTION_CLEANER_CYCLE + "_";
+ try {
+ if (metricsEnabled) {
+ perfLogger.perfLogBegin(AbortedTxnCleaner.class.getName(),
cleanerMetric);
+ }
+ Table t;
+ Partition p = null;
+ t = metadataCache.computeIfAbsent(info.getFullTableName(), () ->
resolveTable(info.dbname, info.tableName));
+ if (isNull(t)) {
+ // The table was dropped before we got around to cleaning it.
+ LOG.info("Unable to find table {}, assuming it was dropped.",
info.getFullTableName());
+ txnHandler.markCleanedForAborts(info);
+ return;
+ }
+ if (MetaStoreUtils.isNoCleanUpSet(t.getParameters())) {
+ // The table was marked no clean up true.
+ LOG.info("Skipping table {} clean up, as NO_CLEANUP set to true",
info.getFullTableName());
+ return;
+ }
+ if (!isNull(info.partName)) {
+ p = resolvePartition(info.dbname, info.tableName, info.partName);
+ if (isNull(p)) {
+ // The partition was dropped before we got around to cleaning it.
+ LOG.info("Unable to find partition {}, assuming it was dropped.",
+ info.getFullPartitionName());
+ txnHandler.markCleanedForAborts(info);
+ return;
+ }
+ if (MetaStoreUtils.isNoCleanUpSet(p.getParameters())) {
+ // The partition was marked no clean up true.
+ LOG.info("Skipping partition {} clean up, as NO_CLEANUP set to
true", info.getFullPartitionName());
+ return;
+ }
+ }
+
+ String location =
CompactorUtil.resolveStorageDescriptor(t,p).getLocation();
+ info.runAs = TxnUtils.findUserToRunAs(location, t, conf);
+ abortCleanUsingAcidDir(info, location, minOpenTxn);
+
+ } catch (Exception e) {
+ LOG.error("Caught exception when cleaning, unable to complete cleaning
of {} due to {}", info,
+ e.getMessage());
+ throw new MetaException(e.getMessage());
+ } finally {
+ if (metricsEnabled) {
+ perfLogger.perfLogEnd(AbortedTxnCleaner.class.getName(),
cleanerMetric);
Review Comment:
@deniskuzZ Yes there is no retry backoff as of now. Even if we want to
implement this, the information related to retry needs to be stored somewhere.
Since we are not using COMPACTION_QUEUE, the only place I see this going is
TXN_COMPONENTS table which I am not sure is desirable or not.
Issue Time Tracking
-------------------
Worklog Id: (was: 857257)
Time Spent: 13h 40m (was: 13.5h)
> Implement a separate handler to handle aborted transaction cleanup
> ------------------------------------------------------------------
>
> Key: HIVE-27020
> URL: https://issues.apache.org/jira/browse/HIVE-27020
> Project: Hive
> Issue Type: Sub-task
> Reporter: Sourabh Badhya
> Assignee: Sourabh Badhya
> Priority: Major
> Labels: pull-request-available
> Time Spent: 13h 40m
> Remaining Estimate: 0h
>
> As described in the parent task, once the cleaner is separated into different
> entities, implement a separate handler which can create requests for aborted
> transactions cleanup. This would move the aborted transaction cleanup
> exclusively to the cleaner.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
