[HADOOP-8689] Make trash a server side configuration option

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.0-alpha
• Fix Version/s: 2.0.2-alpha
• Component/s: fs
• Hadoop Flags: Incompatible change, Reviewed
• Release Note: If fs.trash.interval is configured on the server then the client's value for this configuration is ignored.
• Assignee: Eli Collins
• Reporter: Eli Collins
• Created: 13/Aug/12 22:52
• Updated: 04/Aug/20 03:48
• Resolved: 18/Aug/12 03:25

Description

Per ATM's suggestion in HADOOP-8598 for v2 let's make fs.trash.interval configured server side. If it is not configured server side then the client side configuration is used. The fs.trash.checkpoint.interval option is already server side as the emptier runs in the NameNode. Clients may manually run an emptier via hadoop org.apache.hadoop.fs.Trash but it's OK if it uses a separate interval.

Issue Links

• breaks: HDFS-3856 TestHDFSServerPorts failure is causing surefire fork failure
• is related to: HDFS-15492 Make trash root inside each snapshottable directory
• is related to: ACCUMULO-1254 no-trash option doesn't work against hadoop2 and CDH4
• is related to: HADOOP-8770 NN should not RPC to self to find trash defaults (causes deadlock)
• is related to: HADOOP-8598 Server-side Trash
• is related to: HDFS-4683 Per directory trash settings / trash override
• relates to: HDFS-4889 passing -Dfs.trash.interval to command line is not respected.
• relates to: HADOOP-8710 Remove ability for users to easily run the trash emptier