[HDFS-3955] QJM: Make acceptRecovery() atomic

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: QuorumJournalManager (HDFS-3077)
• Fix Version/s: QuorumJournalManager (HDFS-3077)
• Component/s: ha
• Target Version/s: QuorumJournalManager (HDFS-3077)
• Hadoop Flags: Reviewed
• Assignee: Todd Lipcon
• Reporter: Todd Lipcon
• Created: 19/Sep/12 02:48
• Updated: 19/Sep/12 18:57
• Resolved: 19/Sep/12 18:57

Description

Per one of the TODOs in Journal.java, there is currently a lack of atomicity in the acceptRecovery() code path. In particular, we have the following actions executed non-atomically:

• Download a new edits_inprogress_N from some other node
• Persist the paxos recovery file to disk.

If the JN crashes between these two steps, then we may be left in the state whereby the edits_inprogress file has different data than the Paxos data left over on the disk from a previous recovery attempt. This causes the next prepareRecovery() to fail with an AssertionError.

I discovered this by randomly injecting a fault between the two steps, and then running the randomized fault test on a cluster. This resulted in some AssertionErrors in the test logs.