[
https://issues.apache.org/jira/browse/HIVE-24428?focusedWorklogId=532442&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-532442
]
ASF GitHub Bot logged work on HIVE-24428:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 07/Jan/21 13:02
Start Date: 07/Jan/21 13:02
Worklog Time Spent: 10m
Work Description: deniskuzZ commented on a change in pull request #1724:
URL: https://github.com/apache/hive/pull/1724#discussion_r553313485
##########
File path: ql/src/java/org/apache/hadoop/hive/ql/exec/MoveTask.java
##########
@@ -774,6 +780,100 @@ private void checkFileFormats(Hive db, LoadTableDesc tbd,
Table table)
}
}
+ class LocalTableLock implements Closeable{
+
+ private Optional<HiveLockObject> lock;
+ private HiveLock lockObj;
+
+ public LocalTableLock(Optional<HiveLockObject> lock) throws LockException {
+
+ this.lock = lock;
+ if(!lock.isPresent()) {
+ return;
+ }
+ LOG.info("LocalTableLock; locking: " + lock);
+ HiveLockManager lockMgr = context.getHiveTxnManager().getLockManager();
+ lockObj = lockMgr.lock(lock.get(), HiveLockMode.SEMI_SHARED, true);
+ LOG.info("LocalTableLock; locked: " + lock);
+ }
+
+ @Override
+ public void close() throws IOException {
+ if(!lock.isPresent()) {
+ return;
+ }
+ LOG.info("LocalTableLock; unlocking: "+lock);
+ HiveLockManager lockMgr;
+ try {
+ lockMgr = context.getHiveTxnManager().getLockManager();
+ lockMgr.unlock(lockObj);
+ } catch (LockException e1) {
+ throw new IOException(e1);
+ }
+ LOG.info("LocalTableLock; unlocked");
+ }
+
+ }
+
+ static enum LockFileMoveMode {
+ none, dp, all;
+
+ public static LockFileMoveMode fromConf(HiveConf conf) {
+ if (!conf.getBoolVar(HiveConf.ConfVars.HIVE_SUPPORT_CONCURRENCY)) {
+ return none;
+ }
+ String lockFileMoveMode =
conf.getVar(HiveConf.ConfVars.HIVE_LOCK_FILE_MOVE_MODE);
+ return valueOf(lockFileMoveMode);
+ }
+ }
+
+ private LocalTableLock acquireLockForFileMove(LoadTableDesc loadTableWork)
throws HiveException {
+ LockFileMoveMode mode = LockFileMoveMode.fromConf(conf);
+
+ if (mode == LockFileMoveMode.none) {
+ return new LocalTableLock(Optional.empty());
+ }
+ if (mode == LockFileMoveMode.dp) {
+ if (loadTableWork.getDPCtx() == null) {
+ return new LocalTableLock(Optional.empty());
+ }
+ }
+
+ WriteEntity output = context.getLoadTableOutputMap().get(loadTableWork);
+ List<HiveLockObj> lockObjects = context.getOutputLockObjects().get(output);
+ if (lockObjects == null) {
+ return new LocalTableLock(Optional.empty());
+ }
+ TableDesc table = loadTableWork.getTable();
+ if(table == null) {
+ return new LocalTableLock(Optional.empty());
+ }
+
+ Hive db = getHive();
+ Table baseTable = db.getTable(loadTableWork.getTable().getTableName());
+
+ HiveLockObject.HiveLockObjectData lockData =
+ new HiveLockObject.HiveLockObjectData(queryPlan.getQueryId(),
+ String.valueOf(System.currentTimeMillis()),
+ "IMPLICIT",
+ queryPlan.getQueryStr(),
+ conf);
+
+ HiveLockObject lock = new HiveLockObject(baseTable,lockData);
+
+ for (HiveLockObj hiveLockObj : lockObjects) {
+ if (Arrays.equals(hiveLockObj.getObj().getPaths(), lock.getPaths())) {
+ HiveLockMode l = hiveLockObj.getMode();
+ if (l == HiveLockMode.EXCLUSIVE || l == HiveLockMode.SEMI_SHARED) {
+ // no need to lock ; already owns a more powerful one
+ return new LocalTableLock(Optional.empty());
+ }
+ }
+ }
+
+ return new LocalTableLock(Optional.of(lock));
+ }
+
private boolean isSkewedStoredAsDirs(LoadTableDesc tbd) {
return (tbd.getLbCtx() == null) ? false : tbd.getLbCtx()
Review comment:
minor: tbd.getLbCtx() != null && tbd.getLbCtx().isSkewedStoredAsDir()
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
Issue Time Tracking
-------------------
Worklog Id: (was: 532442)
Time Spent: 3h (was: 2h 50m)
> Concurrent add_partitions requests may lead to data loss
> --------------------------------------------------------
>
> Key: HIVE-24428
> URL: https://issues.apache.org/jira/browse/HIVE-24428
> Project: Hive
> Issue Type: Bug
> Reporter: Zoltan Haindrich
> Assignee: Zoltan Haindrich
> Priority: Major
> Labels: pull-request-available
> Time Spent: 3h
> Remaining Estimate: 0h
>
> in case multiple clients are adding partitions to the same table - when the
> same partition is being added there is a chance that the data dir is removed
> after the other client have already written its data
> https://github.com/apache/hive/blob/5e96b14a2357c66a0640254d5414bc706d8be852/standalone-metastore/metastore-server/src/main/java/org/apache/hadoop/hive/metastore/HiveMetaStore.java#L3958
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
