Skip to content

2.8.2.5
Compare
Choose a tag to compare
@berthoug berthoug released this 11 Jan 15:08
2.8.2.5
21f2391 
    Release notes - hops - Version 2.8.2.5

Bug

  • [HOPS-156] - Fix Hops Block Reporting System to work with multiple namenodes
  • [HOPS-431] - apply HDFS-5312
  • [HOPS-432] - apply HDFS-5629
  • [HOPS-436] - apply HDFS-5661
  • [HOPS-438] - apply HDFS-5540
  • [HOPS-439] - apply HDFS-5691
  • [HOPS-440] - apply HDFS-5690
  • [HOPS-441] - apply HDFS-5675
  • [HOPS-442] - apply HDFS-5582
  • [HOPS-443] - apply HDFS-5671
  • [HOPS-445] - apply HDFS-5449
  • [HOPS-446] - apply HDFS-5756
  • [HOPS-447] - apply HDFS-5747
  • [HOPS-448] - find block by inode id return wrong result when using getIndividualBlockLock
  • [HOPS-450] - apply HDFS-5579
  • [HOPS-451] - apply HDFS-5762
  • [HOPS-452] - apply HDFS-5766
  • [HOPS-454] - apply HDFS-5704
  • [HOPS-455] - apply HDFS-5777
  • [HOPS-456] - apply HDFS-5800
  • [HOPS-461] - HDFS web UI not working with proxy servlet
  • [HOPS-463] - The transaction title for accessing client cache is set to "CONCAT". Update the name to CLIENT_CACHE.
  • [HOPS-466] - apply HDFS-5728
  • [HOPS-468] - Block Report fails to reset hashes of empty buckets on block report
  • [HOPS-471] - apply HDFS-5842
  • [HOPS-472] - apply HDFS-5843
  • [HOPS-473] - apply HDFS-5790
  • [HOPS-474] - apply HDFS-5856
  • [HOPS-477] - Reclaim STO locks set by dead NNs
  • [HOPS-480] - BockInfoDataAccess.existsOnAnyStorage is doing a full table scan
  • [HOPS-481] - INodeFile.getStoragePolicyID need readcommited lock on all parent
  • [HOPS-484] - BlockManager.processMisReplicatedBlocks is too slow
  • [HOPS-485] - add schema update 2.8.2.4 to 2.8.2.5 to the create-tables script
  • [HOPS-487] - Cgroup hierarchy creation does not always work, make improvements and more logging
  • [HOPS-489] - apply HDFS-5859
  • [HOPS-491] - apply HDFS-5791
  • [HOPS-492] - Generate client version of ssl-server.xml for MapReduce/Hive/Tez
  • [HOPS-494] - apply HDFS-5876
  • [HOPS-496] - apply HDFS-5873
  • [HOPS-497] - apply HDFS-5881
  • [HOPS-499] - apply HDFS-5882
  • [HOPS-500] - apply HDFS-5837
  • [HOPS-502] - apply HDFS-4858
  • [HOPS-503] - apply HDFS-5879
  • [HOPS-506] - Fixing subtree operations
  • [HOPS-508] - The lock level of sessions should be set back to read_commited when session are put back in the sessionPooL
  • [HOPS-512] - apply HDFS-5904
  • [HOPS-513] - apply HDFS-5934
  • [HOPS-515] - Apply HDFS-5901
  • [HOPS-517] - apply HDFS-5943
  • [HOPS-518] - apply HDFS-3128
  • [HOPS-519] - apply HDFS-5949
  • [HOPS-520] - UsersGroups.getGroupsFromDB() not committing transaction when user is null
  • [HOPS-521] - apply HDFS-5716
  • [HOPS-522] - apply HDFS-5953
  • [HOPS-523] - apply HDFS-5759
  • [HOPS-524] - apply HDFS-5780
  • [HOPS-525] - apply HDFS-5893
  • [HOPS-549] - Retry mechanism on the client side for failed operations is broken
  • [HOPS-550] - Unsetting the MetaEnabled for a subtree shouldn't write log entries to hdfs_metadata_log
  • [HOPS-552] - Apply HDFS-6057
  • [HOPS-553] - Apply HDFS-6061
  • [HOPS-554] - Apply HDFS-6059
  • [HOPS-555] - Apply HDFS-6067
  • [HOPS-556] - Apply HDFS-6065
  • [HOPS-557] - Apply HDFS-6069
  • [HOPS-558] - Apply HDFS-5986
  • [HOPS-559] - Apply HDFS-6071
  • [HOPS-561] - Apply HDFS-6077
  • [HOPS-564] - Apply HDFS-6079
  • [HOPS-565] - Apply HDFS-5705
  • [HOPS-566] - Apply HDFS-6096
  • [HOPS-567] - Apply HDFS-6097
  • [HOPS-571] - hops-metatadata-dal-impl pom pointing to both kompics and bbc1
  • [HOPS-572] - Script to build native clusterj only works with ndb 7.5
  • [HOPS-574] - apply HDFS-5321
  • [HOPS-575] - apply HDFS-5857
  • [HOPS-576] - apply HDFS-6058
  • [HOPS-577] - apply HDFS-6102
  • [HOPS-578] - apply HDFS-6094
  • [HOPS-579] - apply HDFS-5516
  • [HOPS-581] - apply HDFS-6099
  • [HOPS-582] - apply HDFS-6100
  • [HOPS-583] - apply HDFS-6105
  • [HOPS-584] - apply HDFS-6129
  • [HOPS-586] - apply HDFS-6140
  • [HOPS-587] - apply HDFS-3087
  • [HOPS-589] - apply HDFS-6115
  • [HOPS-590] - apply HDFS-6162
  • [HOPS-592] - apply HDFS-4564
  • [HOPS-594] - apply HDFS-6190
  • [HOPS-595] - apply HDFS-6189
  • [HOPS-596] - apply HDFS-6159
  • [HOPS-597] - apply HDFS-6180
  • [HOPS-599] - apply HDFS-5830
  • [HOPS-601] - apply HDFS-5895
  • [HOPS-602] - apply HDFS-5900
  • [HOPS-603] - compleat application of HDFS-5940 to Cache related code
  • [HOPS-604] - finnish applying HDFS-5950 to cache related code
  • [HOPS-606] - apply HDFS-6086
  • [HOPS-607] - apply HDFS-6106
  • [HOPS-608] - apply HDFS-6107
  • [HOPS-609] - addendum to HOPS-590 to apply HDFS-6162 to cache related code
  • [HOPS-611] - apply HDFS-6143
  • [HOPS-612] - apply HDFS-6160
  • [HOPS-613] - apply HDFS-6206
  • [HOPS-614] - apply HDFS-6209
  • [HOPS-616] - apply HDFS-5669
  • [HOPS-618] - apply HDFS-6231
  • [HOPS-619] - apply HDFS-6232
  • [HOPS-620] - apply HDFS-6237
  • [HOPS-621] - apply HDFS-6208
  • [HOPS-622] - apply HDFS-6234
  • [HOPS-624] - Client fails to initiate a TLS connection to DataNode
  • [HOPS-625] - Node removal too slow
  • [HOPS-626] - Leader election protocol falsely assumes there is a leader
  • [HOPS-627] - replicas stay in the database when datanode removed and namenode restart
  • [HOPS-628] - HOPS-625 not good enough
  • [HOPS-629] - Datanode get allocated block report capacity while not able to block report
  • [HOPS-630] - block reporting should log when failling
  • [HOPS-631] - HOPS-628 fail when number of inodes "present" on datanode is greater than 4096
  • [HOPS-635] - If dfs.namenode.servicerpc-address is not configured services should use dfs.namenode.rpc-address
  • [HOPS-636] - INodeLock.acquireINodeLocks does not work on multiple paths when one of the path does not exist
  • [HOPS-637] - GET_BLOCK_LOCATIONS result in too many logs for LockUpgradeException
  • [HOPS-646] - The balancer create a high load on the database
  • [HOPS-647] - BlockInfoUnderConstruction.initializeBlockRecovery not updating the database when seting chosenAsPrimary to false
  • [HOPS-648] - Do not delete replica table in tests
  • [HOPS-649] - problems with variable table when updating from 2.8.2.4 to 2.8.2.5
  • [HOPS-650] - new NN web ui does not display correctly
  • [HOPS-651] - RetryCacheCleaner use of System.nanoTime is not compatible with distributed NN and restart
  • [HOPS-652] - BlockManager.processMissreplicatedInt is not thread safe
  • [HOPS-653] - FSNameSytem.incrementSafeBlockCount not checking replication
  • [HOPS-654] - Excess replica block metric going negative
  • [HOPS-674] - Use FQDN when authenticating incoming RPCs with CN
  • [HOPS-685] - CRL validation does not work with new keystore schema
  • [HOPS-688] - INodeUtils.getINodeIdsForBlockIds should not try to get too many blocks at once.
  • [HOPS-692] - NameNodeResourceMonitor should check the health of the DB instead of QuotaUpdateMonitor
  • [HOPS-738] - Calls to concurrentBlkReports fail
  • [HOPS-739] - BlockInfo.isReplicatedOnDatanode access the dabase without using the entityManager
  • [HOPS-740] - Call to findBlockInfosByStorageId can overload the database
  • [HOPS-741] - BlockManager.processReport should parallelize adding blocks for first report
  • [HOPS-742] - FSNameSystem.addSafeBlocks can overload the database
  • [HOPS-743] - Datanode removal is too slow
  • [HOPS-746] - Fix client API to get number of namenodes.
  • [HOPS-749] - UnderReplicatedBlockClusterj.removeAll() does not scale
  • [HOPS-809] - Check if retryCache is enabled before acquiring its lock
  • [HOPS-814] - LightWeightRequestHandler should ensure that the session is returned
  • [HOPS-815] - the Slicer should not have its own thread pool
  • [HOPS-816] - Improve randomization in HopsRandomStickyFailoverProxyProvider
  • [HOPS-820] - Startup safe block count is wrong
  • [HOPS-821] - Non leader namenode may never properly leave safe mode
  • [HOPS-825] - Yarn quota service does not handle node manager death properly
  • [HOPS-827] - FSNamesystem.addSafeBlocks can creat deadlock when called in BlockManager.addStoredBlockImmediateTX
  • [HOPS-828] - apply HADOOP-14597
  • [HOPS-867] - Fixing string equality
  • [HOPS-891] - Potential NullPointers in FSNamesystem.getAdditionalDatanode

Sub-task

  • [HOPS-395] - Add version for app certificates
  • [HOPS-398] - Introduce reloadable key managers to HopsSSLSocketFactory
  • [HOPS-416] - Register application with X509 certificate renewer
  • [HOPS-449] - Update crypto material in CertificateLocalizationService
  • [HOPS-460] - Notify RMNodes about new app certificates
  • [HOPS-462] - NodeManagers should update cryptographic material for containers
  • [HOPS-470] - Rebuild UpdatedCryptoMaterial events on RMNode when RM recovers
  • [HOPS-478] - Revoke previous App certificate after renewal
  • [HOPS-486] - Send crypto material version of running apps during NM registration

New Feature

Improvement

Assets 2
Loading