[
https://issues.apache.org/jira/browse/HIVE-26107?focusedWorklogId=763390&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-763390
]
ASF GitHub Bot logged work on HIVE-26107:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 28/Apr/22 09:44
Start Date: 28/Apr/22 09:44
Worklog Time Spent: 10m
Work Description: deniskuzZ commented on code in PR #3172:
URL: https://github.com/apache/hive/pull/3172#discussion_r860693661
##########
ql/src/java/org/apache/hadoop/hive/ql/DriverTxnHandler.java:
##########
@@ -303,8 +303,15 @@ void setWriteIdForAcidFileSinks() throws
SemanticException, LockException {
private void allocateWriteIdForAcidAnalyzeTable() throws LockException {
if (driverContext.getPlan().getAcidAnalyzeTable() != null) {
+ //Inside a compaction transaction, only stats gathering is running which
is not requiring a new write id,
+ //and for duplicate compaction detection it is necessary to not
increment it.
+ boolean isWithinCompactionTxn =
Boolean.parseBoolean(SessionState.get().getHiveVariables().get(Constants.INSIDE_COMPACTION_TRANSACTION_FLAG));
Table table = driverContext.getPlan().getAcidAnalyzeTable().getTable();
- driverContext.getTxnManager().getTableWriteId(table.getDbName(),
table.getTableName());
+ if(isWithinCompactionTxn) {
+
driverContext.getTxnManager().allocateMaxTableWriteId(table.getDbName(),
table.getTableName());
Review Comment:
instead of this could we supply compaction HWM here to avoid db call?
Issue Time Tracking
-------------------
Worklog Id: (was: 763390)
Time Spent: 2h (was: 1h 50m)
> Worker shouldn't inject duplicate entries in `ready for cleaning` state into
> the compaction queue
> -------------------------------------------------------------------------------------------------
>
> Key: HIVE-26107
> URL: https://issues.apache.org/jira/browse/HIVE-26107
> Project: Hive
> Issue Type: Improvement
> Reporter: László Végh
> Assignee: László Végh
> Priority: Major
> Labels: pull-request-available
> Time Spent: 2h
> Remaining Estimate: 0h
>
> How to reproduce:
> 1) create an acid table and load some data ;
> 2) manually trigger the compaction for the table several times;
> 4) inspect compaction_queue: There are multiple entries in 'ready for
> cleaning' state for the same table.
>
> Expected behavior: All compaction request after the first one should be
> rejected until the table is changed again.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
