[HDFS-2693] Synchronization issues around state transition

• Type: Sub-task
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: HA branch (HDFS-1623)
• Fix Version/s: HA branch (HDFS-1623)
• Component/s: ha, namenode
• Target Version/s: HA branch (HDFS-1623)
• Hadoop Flags: Reviewed
• Assignee: Todd Lipcon
• Reporter: Todd Lipcon
• Created: 15/Dec/11 23:18
• Updated: 15/May/20 14:22
• Resolved: 21/Dec/11 03:03

Description

Currently when the NN changes state, it does so without synchronization. In particular, the state transition function does:
(1) leave old state
(2) change state variable
(3) enter new state

This means that the NN is marked as "active" before it has actually transitioned to active mode and opened its edit logs. This gives a window where write transactions can come in and the checkOperation allows them, but then they fail because the edit log is not yet opened.

Issue Links

• blocks: HDFS-1972 HA: Datanode fencing mechanism
• relates to: HDFS-15356 Unify configuration `dfs.ha.allow.stale.reads` to DFSConfigKeys