|
|
@@ -14,6 +14,11 @@ Trunk (unreleased changes)
|
|
|
HDFS-234. Integration with BookKeeper logging system. (Ivan Kelly
|
|
|
via jitendra)
|
|
|
|
|
|
+ HDFS-1623. High Availability Framework for HDFS NN. Contributed by Todd
|
|
|
+ Lipcon, Aaron T. Myers, Eli Collins, Uma Maheswara Rao G, Bikas Saha,
|
|
|
+ Suresh Srinivas, Jitendra Nath Pandey, Hari Mankude, Brandon Li, Sanjay
|
|
|
+ Radia, Mingjie Lai, and Gregory Chanan
|
|
|
+
|
|
|
IMPROVEMENTS
|
|
|
|
|
|
HDFS-1620. Rename HdfsConstants -> HdfsServerConstants, FSConstants ->
|
|
|
@@ -124,6 +129,322 @@ Trunk (unreleased changes)
|
|
|
HDFS-3037. TestMulitipleNNDataBlockScanner#testBlockScannerAfterRestart is
|
|
|
racy. (atm)
|
|
|
|
|
|
+ BREAKDOWN OF HDFS-1623 SUBTASKS
|
|
|
+
|
|
|
+ HDFS-2179. Add fencing framework and mechanisms for NameNode HA. (todd)
|
|
|
+
|
|
|
+ HDFS-1974. Introduce active and standy states to the namenode. (suresh)
|
|
|
+
|
|
|
+ HDFS-2407. getServerDefaults and getStats don't check operation category (atm)
|
|
|
+
|
|
|
+ HDFS-1973. HA: HDFS clients must handle namenode failover and switch over to
|
|
|
+ the new active namenode. (atm)
|
|
|
+
|
|
|
+ HDFS-2301. Start/stop appropriate namenode services when transition to active
|
|
|
+ and standby states. (suresh)
|
|
|
+
|
|
|
+ HDFS-2231. Configuration changes for HA namenode. (suresh)
|
|
|
+
|
|
|
+ HDFS-2418. Change ConfiguredFailoverProxyProvider to take advantage of
|
|
|
+ HDFS-2231. (atm)
|
|
|
+
|
|
|
+ HDFS-2393. Mark appropriate methods of ClientProtocol with the idempotent
|
|
|
+ annotation. (atm)
|
|
|
+
|
|
|
+ HDFS-2523. Small NN fixes to include HAServiceProtocol and prevent NPE on
|
|
|
+ shutdown. (todd)
|
|
|
+
|
|
|
+ HDFS-2577. NN fails to start since it tries to start secret manager in
|
|
|
+ safemode. (todd)
|
|
|
+
|
|
|
+ HDFS-2582. Scope dfs.ha.namenodes config by nameservice (todd)
|
|
|
+
|
|
|
+ HDFS-2591. MiniDFSCluster support to mix and match federation with HA (todd)
|
|
|
+
|
|
|
+ HDFS-1975. Support for sharing the namenode state from active to standby.
|
|
|
+ (jitendra, atm, todd)
|
|
|
+
|
|
|
+ HDFS-1971. Send block report from datanode to both active and standby
|
|
|
+ namenodes. (sanjay, todd via suresh)
|
|
|
+
|
|
|
+ HDFS-2616. Change DatanodeProtocol#sendHeartbeat() to return HeartbeatResponse.
|
|
|
+ (suresh)
|
|
|
+
|
|
|
+ HDFS-2622. Fix TestDFSUpgrade in HA branch. (todd)
|
|
|
+
|
|
|
+ HDFS-2612. Handle refreshNameNodes in federated HA clusters (todd)
|
|
|
+
|
|
|
+ HDFS-2623. Add test case for hot standby capability (todd)
|
|
|
+
|
|
|
+ HDFS-2626. BPOfferService.verifyAndSetNamespaceInfo needs to be synchronized
|
|
|
+ (todd)
|
|
|
+
|
|
|
+ HDFS-2624. ConfiguredFailoverProxyProvider doesn't correctly stop
|
|
|
+ ProtocolTranslators (todd)
|
|
|
+
|
|
|
+ HDFS-2625. TestDfsOverAvroRpc failing after introduction of HeartbeatResponse
|
|
|
+ type (todd)
|
|
|
+
|
|
|
+ HDFS-2627. Determine DN's view of which NN is active based on heartbeat
|
|
|
+ responses (todd)
|
|
|
+
|
|
|
+ HDFS-2634. Standby needs to ingest latest edit logs before transitioning to
|
|
|
+ active (todd)
|
|
|
+
|
|
|
+ HDFS-2671. NN should throw StandbyException in response to RPCs in STANDBY
|
|
|
+ state (todd)
|
|
|
+
|
|
|
+ HDFS-2680. DFSClient should construct failover proxy with exponential backoff
|
|
|
+ (todd)
|
|
|
+
|
|
|
+ HDFS-2683. Authority-based lookup of proxy provider fails if path becomes
|
|
|
+ canonicalized (todd)
|
|
|
+
|
|
|
+ HDFS-2689. HA: BookKeeperEditLogInputStream doesn't implement isInProgress()
|
|
|
+ (atm)
|
|
|
+
|
|
|
+ HDFS-2602. NN should log newly-allocated blocks without losing BlockInfo (atm)
|
|
|
+
|
|
|
+ HDFS-2667. Fix transition from active to standby (todd)
|
|
|
+
|
|
|
+ HDFS-2684. Fix up some failing unit tests on HA branch (todd)
|
|
|
+
|
|
|
+ HDFS-2679. Add interface to query current state to HAServiceProtocol (eli via
|
|
|
+ todd)
|
|
|
+
|
|
|
+ HDFS-2677. Web UI should indicate the NN state. (eli via todd)
|
|
|
+
|
|
|
+ HDFS-2678. When a FailoverProxyProvider is used, DFSClient should not retry
|
|
|
+ connection ten times before failing over (atm via todd)
|
|
|
+
|
|
|
+ HDFS-2682. When a FailoverProxyProvider is used, Client should not retry for 45
|
|
|
+ times if it is timing out to connect to server. (Uma Maheswara Rao G via todd)
|
|
|
+
|
|
|
+ HDFS-2693. Fix synchronization issues around state transition (todd)
|
|
|
+
|
|
|
+ HDFS-1972. Fencing mechanism for block invalidations and replications (todd)
|
|
|
+
|
|
|
+ HDFS-2714. Fix test cases which use standalone FSNamesystems (todd)
|
|
|
+
|
|
|
+ HDFS-2692. Fix bugs related to failover from/into safe mode. (todd)
|
|
|
+
|
|
|
+ HDFS-2716. Configuration needs to allow different dfs.http.addresses for each
|
|
|
+ HA NN (todd)
|
|
|
+
|
|
|
+ HDFS-2720. Fix MiniDFSCluster HA support to work properly on Windows. (Uma
|
|
|
+ Maheswara Rao G via todd)
|
|
|
+
|
|
|
+ HDFS-2291. Allow the StandbyNode to make checkpoints in an HA setup. (todd)
|
|
|
+
|
|
|
+ HDFS-2709. Appropriately handle error conditions in EditLogTailer (atm via
|
|
|
+ todd)
|
|
|
+
|
|
|
+ HDFS-2730. Refactor shared HA-related test code into HATestUtil class (todd)
|
|
|
+
|
|
|
+ HDFS-2762. Fix TestCheckpoint timing out on HA branch. (Uma Maheswara Rao G via
|
|
|
+ todd)
|
|
|
+
|
|
|
+ HDFS-2724. NN web UI can throw NPE after startup, before standby state is
|
|
|
+ entered. (todd)
|
|
|
+
|
|
|
+ HDFS-2753. Fix standby getting stuck in safemode when blocks are written while
|
|
|
+ SBN is down. (Hari Mankude and todd via todd)
|
|
|
+
|
|
|
+ HDFS-2773. Reading edit logs from an earlier version should not leave blocks in
|
|
|
+ under-construction state. (todd)
|
|
|
+
|
|
|
+ HDFS-2775. Fix TestStandbyCheckpoints.testBothNodesInStandbyState failing
|
|
|
+ intermittently. (todd)
|
|
|
+
|
|
|
+ HDFS-2766. Test for case where standby partially reads log and then performs
|
|
|
+ checkpoint. (atm)
|
|
|
+
|
|
|
+ HDFS-2738. FSEditLog.selectinputStreams is reading through in-progress streams
|
|
|
+ even when non-in-progress are requested. (atm)
|
|
|
+
|
|
|
+ HDFS-2789. TestHAAdmin.testFailover is failing (eli)
|
|
|
+
|
|
|
+ HDFS-2747. Entering safe mode after starting SBN can NPE. (Uma Maheswara Rao G
|
|
|
+ via todd)
|
|
|
+
|
|
|
+ HDFS-2772. On transition to active, standby should not swallow ELIE. (atm)
|
|
|
+
|
|
|
+ HDFS-2767. ConfiguredFailoverProxyProvider should support NameNodeProtocol.
|
|
|
+ (Uma Maheswara Rao G via todd)
|
|
|
+
|
|
|
+ HDFS-2795. Standby NN takes a long time to recover from a dead DN starting up.
|
|
|
+ (todd)
|
|
|
+
|
|
|
+ HDFS-2592. Balancer support for HA namenodes. (Uma Maheswara Rao G via todd)
|
|
|
+
|
|
|
+ HDFS-2367. Enable the configuration of multiple HA cluster addresses. (atm)
|
|
|
+
|
|
|
+ HDFS-2812. When becoming active, the NN should treat all leases as freshly
|
|
|
+ renewed. (todd)
|
|
|
+
|
|
|
+ HDFS-2737. Automatically trigger log rolls periodically on the active NN. (todd
|
|
|
+ and atm)
|
|
|
+
|
|
|
+ HDFS-2820. Add a simple sanity check for HA config (todd)
|
|
|
+
|
|
|
+ HDFS-2688. Add tests for quota tracking in an HA cluster. (todd)
|
|
|
+
|
|
|
+ HDFS-2804. Should not mark blocks under-replicated when exiting safemode (todd)
|
|
|
+
|
|
|
+ HDFS-2807. Service level authorizartion for HAServiceProtocol. (jitendra)
|
|
|
+
|
|
|
+ HDFS-2809. Add test to verify that delegation tokens are honored after
|
|
|
+ failover. (jitendra and atm)
|
|
|
+
|
|
|
+ HDFS-2838. NPE in FSNamesystem when in safe mode. (Gregory Chanan via eli)
|
|
|
+
|
|
|
+ HDFS-2805. Add a test for a federated cluster with HA NNs. (Brandon Li via
|
|
|
+ jitendra)
|
|
|
+
|
|
|
+ HDFS-2841. HAAdmin does not work if security is enabled. (atm)
|
|
|
+
|
|
|
+ HDFS-2691. Fixes for pipeline recovery in an HA cluster: report RBW replicas
|
|
|
+ immediately upon pipeline creation. (todd)
|
|
|
+
|
|
|
+ HDFS-2824. Fix failover when prior NN died just after creating an edit log
|
|
|
+ segment. (atm via todd)
|
|
|
+
|
|
|
+ HDFS-2853. HA: NN fails to start if the shared edits dir is marked required
|
|
|
+ (atm via eli)
|
|
|
+
|
|
|
+ HDFS-2845. SBN should not allow browsing of the file system via web UI. (Bikas
|
|
|
+ Saha via atm)
|
|
|
+
|
|
|
+ HDFS-2742. HA: observed dataloss in replication stress test. (todd via eli)
|
|
|
+
|
|
|
+ HDFS-2870. Fix log level for block debug info in processMisReplicatedBlocks
|
|
|
+ (todd)
|
|
|
+
|
|
|
+ HDFS-2859. LOCAL_ADDRESS_MATCHER.match has NPE when called from
|
|
|
+ DFSUtil.getSuffixIDs when the host is incorrect (Bikas Saha via todd)
|
|
|
+
|
|
|
+ HDFS-2861. checkpointing should verify that the dfs.http.address has been
|
|
|
+ configured to a non-loopback for peer NN (todd)
|
|
|
+
|
|
|
+ HDFS-2860. TestDFSRollback#testRollback is failing. (atm)
|
|
|
+
|
|
|
+ HDFS-2769. HA: When HA is enabled with a shared edits dir, that dir should be
|
|
|
+ marked required. (atm via eli)
|
|
|
+
|
|
|
+ HDFS-2863. Failures observed if dfs.edits.dir and shared.edits.dir have same
|
|
|
+ directories. (Bikas Saha via atm)
|
|
|
+
|
|
|
+ HDFS-2874. Edit log should log to shared dirs before local dirs. (todd)
|
|
|
+
|
|
|
+ HDFS-2890. DFSUtil#getSuffixIDs should skip unset configurations. (atm)
|
|
|
+
|
|
|
+ HDFS-2792. Make fsck work. (atm)
|
|
|
+
|
|
|
+ HDFS-2808. HA: haadmin should use namenode ids. (eli)
|
|
|
+
|
|
|
+ HDFS-2819. Document new HA-related configs in hdfs-default.xml. (eli)
|
|
|
+
|
|
|
+ HDFS-2752. HA: exit if multiple shared dirs are configured. (eli)
|
|
|
+
|
|
|
+ HDFS-2894. HA: automatically determine the nameservice Id if only one
|
|
|
+ nameservice is configured. (eli)
|
|
|
+
|
|
|
+ HDFS-2733. Document HA configuration and CLI. (atm)
|
|
|
+
|
|
|
+ HDFS-2794. Active NN may purge edit log files before standby NN has a chance to
|
|
|
+ read them (todd)
|
|
|
+
|
|
|
+ HDFS-2901. Improvements for SBN web UI - not show under-replicated/missing
|
|
|
+ blocks. (Brandon Li via jitendra)
|
|
|
+
|
|
|
+ HDFS-2905. HA: Standby NN NPE when shared edits dir is deleted. (Bikas Saha via
|
|
|
+ jitendra)
|
|
|
+
|
|
|
+ HDFS-2579. Starting delegation token manager during safemode fails. (todd)
|
|
|
+
|
|
|
+ HDFS-2510. Add HA-related metrics. (atm)
|
|
|
+
|
|
|
+ HDFS-2924. Standby checkpointing fails to authenticate in secure cluster.
|
|
|
+ (todd)
|
|
|
+
|
|
|
+ HDFS-2915. HA: TestFailureOfSharedDir.testFailureOfSharedDir() has race
|
|
|
+ condition. (Bikas Saha via jitendra)
|
|
|
+
|
|
|
+ HDFS-2912. Namenode not shutting down when shared edits dir is inaccessible.
|
|
|
+ (Bikas Saha via atm)
|
|
|
+
|
|
|
+ HDFS-2917. HA: haadmin should not work if run by regular user (eli)
|
|
|
+
|
|
|
+ HDFS-2939. TestHAStateTransitions fails on Windows. (Uma Maheswara Rao G via
|
|
|
+ atm)
|
|
|
+
|
|
|
+ HDFS-2947. On startup NN throws an NPE in the metrics system. (atm)
|
|
|
+
|
|
|
+ HDFS-2942. TestActiveStandbyElectorRealZK fails if build dir does not exist.
|
|
|
+ (atm)
|
|
|
+
|
|
|
+ HDFS-2948. NN throws NPE during shutdown if it fails to startup (todd)
|
|
|
+
|
|
|
+ HDFS-2909. HA: Inaccessible shared edits dir not getting removed from FSImage
|
|
|
+ storage dirs upon error. (Bikas Saha via jitendra)
|
|
|
+
|
|
|
+ HDFS-2934. Allow configs to be scoped to all NNs in the nameservice. (todd)
|
|
|
+
|
|
|
+ HDFS-2935. Shared edits dir property should be suffixed with nameservice and
|
|
|
+ namenodeID (todd)
|
|
|
+
|
|
|
+ HDFS-2928. ConfiguredFailoverProxyProvider should not create a NameNode proxy
|
|
|
+ with an underlying retry proxy. (Uma Maheswara Rao G via atm)
|
|
|
+
|
|
|
+ HDFS-2955. IllegalStateException during standby startup in getCurSegmentTxId.
|
|
|
+ (Hari Mankude via atm)
|
|
|
+
|
|
|
+ HDFS-2937. TestDFSHAAdmin needs tests with MiniDFSCluster. (Brandon Li via
|
|
|
+ suresh)
|
|
|
+
|
|
|
+ HDFS-2586. Add protobuf service and implementation for HAServiceProtocol.
|
|
|
+ (suresh via atm)
|
|
|
+
|
|
|
+ HDFS-2952. NN should not start with upgrade option or with a pending an
|
|
|
+ unfinalized upgrade. (atm)
|
|
|
+
|
|
|
+ HDFS-2974. MiniDFSCluster does not delete standby NN name dirs during format.
|
|
|
+ (atm)
|
|
|
+
|
|
|
+ HDFS-2929. Stress test and fixes for block synchronization (todd)
|
|
|
+
|
|
|
+ HDFS-2972. Small optimization building incremental block report (todd)
|
|
|
+
|
|
|
+ HDFS-2973. Re-enable NO_ACK optimization for block deletion. (todd)
|
|
|
+
|
|
|
+ HDFS-2922. HA: close out operation categories (eli)
|
|
|
+
|
|
|
+ HDFS-2993. HA: BackupNode#checkOperation should permit CHECKPOINT operations
|
|
|
+ (eli)
|
|
|
+
|
|
|
+ HDFS-2904. Client support for getting delegation tokens. (todd)
|
|
|
+
|
|
|
+ HDFS-3013. HA: NameNode format doesn't pick up
|
|
|
+ dfs.namenode.name.dir.NameServiceId configuration (Mingjie Lai via todd)
|
|
|
+
|
|
|
+ HDFS-3019. Fix silent failure of TestEditLogJournalFailures (todd)
|
|
|
+
|
|
|
+ HDFS-2958. Sweep for remaining proxy construction which doesn't go through
|
|
|
+ failover path. (atm)
|
|
|
+
|
|
|
+ HDFS-2920. fix remaining TODO items. (atm and todd)
|
|
|
+
|
|
|
+ HDFS-3027. Implement a simple NN health check. (atm)
|
|
|
+
|
|
|
+ HDFS-3023. Optimize entries in edits log for persistBlocks call. (todd)
|
|
|
+
|
|
|
+ HDFS-2979. Balancer should use logical uri for creating failover proxy with HA
|
|
|
+ enabled. (atm)
|
|
|
+
|
|
|
+ HDFS-3035. Fix failure of TestFileAppendRestart due to OP_UPDATE_BLOCKS (todd)
|
|
|
+
|
|
|
+ HDFS-3039. Address findbugs and javadoc warnings on branch. (todd via atm)
|
|
|
+
|
|
|
Release 0.23.3 - UNRELEASED
|
|
|
|
|
|
INCOMPATIBLE CHANGES
|
Aaron Myers