[HADOOP-16479] ABFS FileStatus.getModificationTime returns localized time instead of UTC

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.2.0
• Fix Version/s: 3.3.0
• Component/s: fs/azure

Release Note

private long parseLastModifiedTime(final String lastModifiedTime)

Timezone is part of lastModifiedTime String as it's last field. But when parsed it ignores timezone field and assumes JVM timezone.
Fix: Made timezone field considered in lastModifiedTime when the same is parsed

• Assignee: Bilahari T H
• Reporter: Joan Sala Reixach
• Created: 31/Jul/19 16:28
• Updated: 16/Apr/21 15:03
• Resolved: 08/Aug/19 18:10

Description

As per javadoc, the method FileStatus.getModificationTime() should return the time in UTC, but it returns the time in the JVM timezone.

The issue origins in AzureBlobFileSystemStore.getFileStatus() itself, since  parseLastModifiedTime() returns a wrong date. I have created a file in Azure Data Lake Gen2 and when I look at  it through the Azure Explorer it shows the correct modification time, but the method returns -2 hours time (I am in CET = UTC+2).

Azure Explorer last modified time:

AbfsClient parseLastModifiedTime:

It shows 15:21 CEST as utcDate, when it should be 15:21 UTC, which results in the 2 hour loss.

DateFormat.parse uses a localized calendar to parse dates which might be the source of the issue.

Issue Links

• is related to: HADOOP-17640 ABFS: transient failure of TestAzureBlobFileSystemFileStatus.testLastModifiedTime
• links to: GitHub Pull Request #1207