[
https://issues.apache.org/jira/browse/GOBBLIN-1905?focusedWorklogId=880258&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-880258
]
ASF GitHub Bot logged work on GOBBLIN-1905:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 13/Sep/23 19:07
Start Date: 13/Sep/23 19:07
Worklog Time Spent: 10m
Work Description: ZihanLi58 commented on code in PR #3769:
URL: https://github.com/apache/gobblin/pull/3769#discussion_r1324950900
##########
gobblin-modules/gobblin-orc/src/main/java/org/apache/gobblin/writer/GobblinBaseOrcWriter.java:
##########
@@ -211,6 +211,9 @@ public void flush()
initializeOrcFileWriter();
}
orcFileWriter.addRowBatch(rowBatch);
+ if (this.selfTuningWriter) {
+ this.currentOrcWriterMaxUnderlyingMemory =
Math.max(this.currentOrcWriterMaxUnderlyingMemory,
orcFileWriter.estimateMemory());
Review Comment:
Add one comment here to mention that reset will reduce the fileWriter's
memory?
Issue Time Tracking
-------------------
Worklog Id: (was: 880258)
Time Spent: 20m (was: 10m)
> Fix bug where ORC self tune would under estimate native ORC writer memory
> -------------------------------------------------------------------------
>
> Key: GOBBLIN-1905
> URL: https://issues.apache.org/jira/browse/GOBBLIN-1905
> Project: Apache Gobblin
> Issue Type: Bug
> Components: gobblin-core
> Reporter: William Lo
> Assignee: Abhishek Tiwari
> Priority: Major
> Time Spent: 20m
> Remaining Estimate: 0h
>
> During close, the native ORC writer would flush itself - emptying its buffers
> and leading to the memory estimate to be extremely minimal, then record its
> memory estimate in the saved state if it has not recorded an estimate before.
> So for datasets that write a small number of records to file it would greatly
> underestimate the memory within the native ORC writer.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
