[HDFS-4258] Rename of Being Written Files

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.0.0-alpha1
• Component/s: hdfs-client, namenode
• Assignee: Brandon Li
• Reporter: Tsz-wo Sze
• Created: 01/Dec/12 23:23
• Updated: 12/May/16 18:13
• Resolved: 12/May/14 18:20

Description

When a being written file or it's ancestor directories is renamed, the path in the file lease is also renamed. Then the writer of the file usually will fail since the file path in the writer is not updated.

Moreover, I think there is a bug as follow:

1. Client writes 0's to F_0="/foo/file" and writes 1's to F_1="/bar/file" at the same time.
2. Rename /bar to /baz
3. Rename /foo to /bar

Then, writing to F_0 will fail since /foo/file does not exist anymore but writing to F_1 may succeed since /bar/file exits as a different file. In such case, the content of /bar/file could be partly 0's and partly 1's.

Issue Links

• duplicates: HDFS-6294 Use INode IDs to avoid conflicts when a file open for write is renamed
• is depended upon by: HADOOP-9361 Strictly define the expected behavior of filesystem APIs and write tests to verify compliance
• relates to: HDFS-4248 Renames may remove file leases
• relates to: HDFS-4489 Use InodeID as as an identifier of a file in HDFS protocols and APIs
• relates to: HDFS-6294 Use INode IDs to avoid conflicts when a file open for write is renamed