hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (Jira)" <j...@apache.org>
Subject [jira] [Work logged] (HIVE-24145) Fix preemption issues in reducers and file sink operators
Date Fri, 11 Sep 2020 05:46:00 GMT

     [ https://issues.apache.org/jira/browse/HIVE-24145?focusedWorklogId=481902&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-481902
]

ASF GitHub Bot logged work on HIVE-24145:
-----------------------------------------

                Author: ASF GitHub Bot
            Created on: 11/Sep/20 05:45
            Start Date: 11/Sep/20 05:45
    Worklog Time Spent: 10m 
      Work Description: rbalamohan commented on a change in pull request #1485:
URL: https://github.com/apache/hive/pull/1485#discussion_r486786544



##########
File path: ql/src/java/org/apache/hadoop/hive/ql/exec/FileSinkOperator.java
##########
@@ -216,29 +216,47 @@ public FSPaths(Path specPath, boolean isMmTable, boolean isDirectInsert,
boolean
     }
 
     public void closeWriters(boolean abort) throws HiveException {
+      Exception exception = null;
       for (int idx = 0; idx < outWriters.length; idx++) {
         if (outWriters[idx] != null) {
           try {
             outWriters[idx].close(abort);
             updateProgress();
           } catch (IOException e) {
-            throw new HiveException(e);
+            exception = e;
+            LOG.error("Error closing " + outWriters[idx].toString(), e);
+            // continue closing others
           }
         }
       }
-      try {
+      for (int i = 0; i < updaters.length; i++) {
+        if (updaters[i] != null) {
+          SerDeStats stats = updaters[i].getStats();
+          // Ignore 0 row files except in case of insert overwrite
+          if (isDirectInsert && (stats.getRowCount() > 0 || isInsertOverwrite))
{
+            outPathsCommitted[i] = updaters[i].getUpdatedFilePath();
+          }
+          try {
+            updaters[i].close(abort);
+          } catch (IOException e) {
+            exception = e;
+            LOG.error("Error closing " + updaters[i].toString(), e);
+            // continue closing others
+          }
+        }
+      }
+      // Made an attempt to close all writers.
+      if (exception != null) {
         for (int i = 0; i < updaters.length; i++) {
           if (updaters[i] != null) {
-            SerDeStats stats = updaters[i].getStats();
-            // Ignore 0 row files except in case of insert overwrite
-            if (isDirectInsert && (stats.getRowCount() > 0 || isInsertOverwrite))
{
-              outPathsCommitted[i] = updaters[i].getUpdatedFilePath();
+            try {
+              fs.delete(updaters[i].getUpdatedFilePath(), true);
+            } catch (IOException e) {
+              e.printStackTrace();

Review comment:
       LOG?




----------------------------------------------------------------
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:
users@infra.apache.org


Issue Time Tracking
-------------------

    Worklog Id:     (was: 481902)
    Time Spent: 20m  (was: 10m)

> Fix preemption issues in reducers and file sink operators
> ---------------------------------------------------------
>
>                 Key: HIVE-24145
>                 URL: https://issues.apache.org/jira/browse/HIVE-24145
>             Project: Hive
>          Issue Type: Bug
>            Reporter: Ramesh Kumar Thangarajan
>            Assignee: Ramesh Kumar Thangarajan
>            Priority: Major
>              Labels: pull-request-available
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> There are two issues because of preemption:
>  # Reducers are gettingĀ reordered as part of optimizations because of which more preemption
happen
>  # Preemption in the middle of writing can cause the file to not close and lead to errors
when we read the file later



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message