Steve Loughran created HADOOP-18004:
---------------------------------------
Summary: abfs and s3a disk buffer factories to use UUIDs for file
prefixes
Key: HADOOP-18004
URL: https://issues.apache.org/jira/browse/HADOOP-18004
Project: Hadoop Common
Issue Type: Improvement
Components: fs/azure, fs/s3
Affects Versions: 3.3.2
Reporter: Steve Loughran
Assignee: Mehakmeet Singh
the disk buffers created in s3a and abfs output streams use a simple
String.format("datablock-%04d-", index) pattern for the prefix for File.tmpFile
this means there will be contention for filenames across streams, especially
across processes.
if each stream had a uuid prefix there'd be no contention. That'd change the
API though. Alternatively: each disk block factory has the uuid, and the index
is simply total number blocks created.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-dev-h...@hadoop.apache.org
