[HADOOP-16490] Avoid/handle cached 404s during S3A file creation

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.3.0
• Fix Version/s: 3.3.0
• Component/s: fs/s3
• Assignee: Steve Loughran
• Reporter: Steve Loughran
• Created: 05/Aug/19 15:36
• Updated: 21/Sep/20 12:48
• Resolved: 11/Sep/19 16:03

Description

If S3Guard is encountering delayed consistency (FNFE from tombstone; failure to open file) then

• it only retries with the same times as everything else. We should make it differently configurable
• when an FNFE is finally thrown, rename() treats it as being caused by the original source path missing, when in fact its something else. Proposed: somehow propagate the failure up differently, probably in the S3AFileSystem.copyFile() code
• don't do HEAD checks when creating files
• shell commands to avoid deleteOnExit calls as these also generate HEAD calls by way of exists() checks

eliminating the HEAD checks will stop 404s getting into the S3 load balancer/cache during file creation

Issue Links

• blocks: HADOOP-14936 S3Guard: remove "experimental" from documentation
• causes: HADOOP-16885 Encryption zone file copy failure leaks temp file ._COPYING_ and wrapped stream
• contains: HADOOP-16280 S3Guard: Retry failed read with backoff in Authoritative mode when file can be opened
• contains: HADOOP-13884 s3a create(overwrite=true) to only look for dir/ and list entries, not file
• contains: HADOOP-16501 s3guard auth path checks only check against unqualified source path
• is duplicated by: HADOOP-17216 delta.io spark task commit encountering S3 cached 404/FileNotFoundException
• is related to: HADOOP-17216 delta.io spark task commit encountering S3 cached 404/FileNotFoundException
• relates to: HADOOP-16499 S3A retry policy to be exponential
• supercedes: HADOOP-15460 S3A FS to add "fs.s3a.create.performance" to the builder file creation option set
• links to: GitHub Pull Request #1229
• links to: GitHub Pull Request #1407