[HDFS-3399] BookKeeper option support for NN HA

• Type: New Feature
• Status: Reopened
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.0.0-alpha, 3.0.0-alpha1
• Component/s: ha
• Target Version/s: 2.0.0-alpha
• Hadoop Flags: Reviewed
• Assignee: Unassigned
• Reporter: Uma Maheswara Rao G
• Created: 10/May/12 17:23
• Updated: 07/Oct/16 05:34

Description

Here is the JIRA to BookKeeper support issues with NN HA. We can file all the BookKeeperJournalManager issues under this JIRA for more easy tracking.

Issue Links

• depends upon: HDFS-234 Integration with BookKeeper logging system
• is related to: HDFS-1623 High Availability Framework for HDFS NN

Sub-Tasks

• BK JM : Namenode is not deleting his lock entry '/ledgers/lock/lock-0000X', when fails to acquire the lock: Resolved
• BookKeeper Journal Manager is not retrying to connect to BK when BookKeeper is not available for write.: Resolved
• Document the BKJM usage in Namenode HA.: Closed
• BKJM : Namenode format fails, if there is no BK root: Closed
• HA: Bring BookKeeperJournalManager up to date with HA changes: Closed
• BookKeeper Journal output stream doesn't check addComplete rc: Closed
• BKJM: NN startup is failing, when tries to recoverUnfinalizedSegments() a bad inProgress_ ZNodes: Closed
• Race condition between rolling logs at active NN and purging at standby: Closed
• BKJM:Switch from standby to active fails and NN gets shut down due to delay in clearing of lock: Closed
• BKJM: Deleting currentLedger and leaving 'inprogress_x' on exceptions can throw BKNoSuchLedgerExistsException later.: Resolved
• Make BKJM-ZK session timeout configurable.: Closed
• Cleanup Exception handling in BookKeeper journal manager: Closed
• Handle disconnect and session timeout events at BKJM: Resolved
• BKJM: zkLatchWaitTimeout hard coded to 6000. Make use of ZKSessionTimeout instead.: Closed
• BOOTSTRAPSTANDBY for new Standby node will not work just after saveNameSpace at ANN in case of BKJM: Resolved
• Make BKJM use protobufs for all serialization with ZK: Closed
• Implement format() for BKJM: Closed
• Implement a fence method which should fence the BK shared storage.: Resolved
• Override toString() for BookKeeperEditLogInputStream: Closed
• standby namenode become active fails because starting log segment fail on shared storage: Closed
• In HA mode, when there is a ledger in BK missing, which is generated after the last checkpoint, NN can not restore it.: Resolved
• BKJM: The reading for editlog at NN starting using bkjm is not efficient: Closed
• BKJM: Two namenodes usng bkjm can race to create the version znode: Resolved
• Remove the format limitation point from BKJM documentation as HDFS-3810 closed: Closed
• BKJM doesn't take advantage of speculative reads: Closed
• BKJM: Separate write and ack quorum: Closed
• Changes from BOOKKEEPER-203 broken capability of including bookkeeper-server jar in hidden package of BKJM: Closed
• All BKJM ledgers are not checked while tailing, So failover will fail.: Closed
• Improve BKJM documentation: Resolved
• TestBookKeeperHACheckpoints fails: Closed
• Update Bookkeeper dependency to 4.2.3: Closed