https://github.com/voldemort/voldemort
Tip revision: 6809fceb924adf60919aee45dceeeb19f520a27c authored by Sidian Wu on 30 March 2016, 00:12:13 UTC
Releasing Voldemort 1.10.14
Releasing Voldemort 1.10.14
Tip revision: 6809fce
release_notes.txt
Release 1.10.14 3/29/2016
* Report metrics for Scheduler and Async Service
* better error message thrown when pushing to a store with storage quota 0
* Resolved concurrency push conflict
* Added a configurable flag on the BnP side that can disable store creation
* Fixed build.gradle
* Fixed a NPE thrown by the StorageEngineService on startup with views
* Added server state checking when bringing it back
* Fixed memory leaking issue in Hdfs FileSystem when fetching
* AdminClient updates version after updating the data
* Added second checksum validation when receiving data
* Added more more aggregatedmetrics for HDFS data
Release 1.10.13 on 2/8/2016
* Unit Test fixes for ClientRequestExecutorPool
* ShadowJar with Protobuf dependency
* HdfsFetcher error message cleanup
Release 1.10.12 on 1/11/2016
* Fixed and cleaned up HadoopStoreWriterTest.
* Fixed HDFS directory size measurement in HdfsFetcher.
Release 1.10.11 on 1/8/2016
* Fixed RO data directory diverged on 1.10.7
* Fixed BnP assigns keys into correct partitions, resulting in data loss
* Fixed unknown.host unit test failure
* Fixed HadoopStoreWriter for multiple chunks bug
* Added Defensive code for BnP Partition reducer when it sees incorrect data
Release 1.10.10 on 1/6/2016
* AdminClient constructor cleanup and respects the timeout of ClientConfig
Release 1.10.9 on 1/5/2016
* Fixed a regression affecting Read-Only servers introduced in 1.10.7.
* Fixed a unit test which failed on Java 8.
* Made vadmin.sh more resilient to node failures.
Release 1.10.8 on 1/4/2016
* Validate compression when a store is pushed via BnP
* BnP HA pushes fail, when fetch is disabled on a node.
* Enhanced Meta check which compares the file name and sizes
* Wait for Scheduler Service to shutdown, before proceeding with shutdown
* JMX errors during register/unregister for RO Stores
Release 1.10.7 on 12/15/2015
* Introduced 'build.primary.replicas.only' mode in BnP.
* Performance tool can now load custom workloads from external files.
* Fixed a bug where the server would fail to start if Bouncy Castle is
absent from the classpath.
* Admin operations improvements:
* Better error message when server receives unknown admin operation.
* Improved the Meta Check operation's default behavior.
* BnP fetches now abort immediately when their async operation is killed.
Release 1.10.6 on 12/08/2015
* BnP loggin improvements
* Option to use BouncyCastle for encryption/decryption, instead of JCE
* JMX Metrics for tracking Server state
* Enable BnP HA only on Connection errors
Release 1.10.5 on 11/25/2015
* Extended voldemort shell to let "preflist" command show parition information.
* Bug fix on python script used to generate cluster.xml.
* Bug fix updating metadata version in cluster.xml and stores.xml.
* Read-only improvements and fixes:
* Add replace URL of feching file to Voldemort node.
Release 1.10.4 on 11/11/2015
* The AdminClient in the VoldemortClientShell is now configurable.
* Improved error messages in BnP and Read-Only server-side code.
* Fixed a server-side bug in BnP HA.
* Made BnP jobs fail explicitly when clusters are inconsistent.
Release 1.10.3 on 11/04/2015
* Bug fix and improvement to the Forklift tool.
* Improved help prompt and configurability of the VoldemortClientShell.
* Moved BnP's verifyOrAddStore() to the AdminClient.
* Implement the truncate function in the RocksDB storage engine.
* N.B.: Previous data is backwards incompatible, see a0e19c6 for details.
Release 1.10.2 on 10/30/2015
* Added a ReadOnlyFileValidator script.
* Fixed rename operation in HdfsFailedFetchLock (regression fix).
Release 1.10.1 on 10/27/2015
* Read-Only server improvements and fixes:
* Read-Only server now avoids needlessly recreating symlinks.
* Individual file checksums now outputted to stats file.
* Removed a harmful set quota admin op from BnP (regression fix).
* Added more details to an exception shown in BnP (regression fix).
Release 1.10.0 on 10/19/2015
* Many many many fixes and improvements in the 1.9.x series of release! Many thanks to all contributors!
* Minor improvements since the last patch release (1.9.22):
* Upgraded RocksDB to 3.13.1 and added more configurability options for it.
* Improvements to the benchmarking tool.
* Added a unit test for colliding keys in Read-Only data files.
* Added various safe-guards in case Read-Only index/data files become corrupted.
Release 1.9.22 on 10/08/2015
* Disk quota for Voldemort Read-Only pipeline
* Improved debuggability of BnP HA
* Added heart beat for NIO selector
* Added additional keepalive settings to avoid client connection leaks
* Trimmed fat from HadoopUtils
* Proper interruption of BnP hooks
Release 1.9.21 on 09/24/2015
* Kerberos token caching now considered experimental (disabled by default).
Release 1.9.20 on 09/14/2015
* BnP/RO improvements:
* Various debuggability improvements.
* Graceful recovery from incomplete store creation.
* HA locking now happens on the server-side.
* Removed 'node.id' BnP config which caused a SPoF.
* Script improvements:
* voldemort-shell.sh properly print Avro data.
* Forklifting makes a proper schema check.
* Binary fetch-entries format.
* Unit tests improvements:
* Longer retry times on ServerTestUtils.startVoldemortServer().
* Ignored failing RocksDB test since it is unimplemented.
* New forklifting tests.
* RocksDB iteration logic is now properly implemented.
* Improved JMX beans management.
* Fixed the routing strategy of the voldsys$_store_quotas store.
Release 1.9.19 on 08/12/2015
* Removed unused things (scala shell, ec2-testing, public-lib directory)
* Fixed some unit tests.
* Various BnP improvements:
* Improved logging.
* Added bnpJar gradle target for bundling BnP with its dependencies.
* Added run-bnp.sh script for running without Azkaban.
* Various Read-Only server-side improvements:
* Upgrade to Hadoop 2.
* Kerberos authentication clean ups.
* Got rid of unused File Descriptor and other MemLock clean ups.
* Added STORAGE_SPACE quota API (not enforced yet).
Release 1.9.18 on 07/20/2015
* Various BnP improvements:
* Introduces of a limited form of High Availability for BnP.
* Improves path handling and validation in VoldemortSwapJob.
* Fetch throttling now works with block-level compression.
* Pushes to multiple colos now happen in parallel.
* Some code refactoring and clean ups.
* Minor bug fixes in scripts:
* Fixes a SecurityException when running HadoopStoreJobRunner in Oozie.
* Fixes Coordinator start script.
* EventThrottler code now uses Tehuti.
Release 1.9.17 on 06/12/2015
* Robust handling of connection failures
* Flush the queue on marking a node down
* Async connect exceptions are remembered based on connection timeout
Release 1.9.16 on 06/09/2015
* Fixes unregistration of JMX mbean.
* Fixes VoldemortConfig bug introduced in 1.9.15.
* Fixes error reporting in AvroUtils.getSchemaFromPath().
* Caches the result of duplicate AvroUtils and HadoopUtils calls in BnP.
Release 1.9.15 on 06/05/2015
* Unit tests for Serialization
* Alert if BnP jobs does not progress
* Add Timeout for ConfigurableSocketStoreFactory
* HdfsCopyStats outputs object ids, not Hdfs Paths
* HdfsCopyStatsTest fails intermittently
Release 1.9.14 on 05/22/2015
* RO stats feature
* HdfsFetcher refactoring
* Hadoop Core and IO commons version upgrade
* RO fetcher GC performance reduction by 90%
* Fix quota redistribution on zone shrinkage issue
* Admin command for meta get ro
* Store delete command fix
* Unit test fixes
Release 1.9.13 on 05/01/2015
* Non blocking connect and Non blocking Protocol negotiation.
* Suppress logging ObsoleteVersionExceptions in request handler.
* Java minimum version is bumped to Java 6.
Release 1.9.12 on 05/01/2015
* quota reset on rebalance
* Fully disabled ant build
* fix rebalance unit test intermittent failures
* Changing buffer size of GZIP streams
* Additional test cases
Release 1.9.11 on 04/23/2015
* BuildAndPush job produces compressed files as dictated by Server
* Voldemort RO Server supports on the fly decompression when compression is
enabled
* Admin request to query Server on supported compression codec
Release 1.9.10 on 04/15/2015
* Share Read/Write buffers for Client
* Share Read/Write buffers for Server on Client Operations
* Separate Client and Admin Request Handlers
* Remove extra allocations on Server side while parsing the request
* Pre-allocate Put request buffer before writing the request
* Remove one extra byte array allocation on Put, when parsing
Vector clocks from the Value
Release 1.9.9 on 04/10/2015
* Admin tools improvements and bug fixes:
* vadmin.sh now deals gracefully with inconsistent metadata and other issues.
* ReplaceNodeCLI now deals gracefully with empty stores.xml.
* ReadOnlyReplicationHelperCLI can now act on the local node, regardless of server version.
* HdfsFetcher imprpvements:
* Can fetch individual files to specific local dir (when used from CLI).
* Lower memory consumption.
* BnP improvements:
* Added a min.number.of.records config (defaults to 1) to prevent pushing empty stores.
* Improved error handling and reporting in BnP's run function.
* Client-related automated tests expanded and improved.
* Metadata queries no longer always hit the same node.
* Initial release of the RocksDB storage engine! (N.B.: still considered experimental).
Release 1.9.8 on 03/17/2015
* Fixes a bug in the vadmin restore-from-replica command.
* Improves the vadmin meta check command.
* Improves the build's tar target.
* Adds better logging in the expansion code.
* Cleans up BnP:
* Removes Azkaban dependency as much as possible.
* Standardizes on using voldemort.utils.Props as much as possible and adds more utility functions to it.
* Deletes VoldemortMultiStoreBuildAndPushJob which is not actively used and suffering from code rot.
* Adds Content-Length header support in BnP HttpHook.
* Adds safeguards in BnP HttpHook's concurrent code.
* Removes System.exit calls from BnP.
* Cleans up dependencies:
* Removed tusk, RestHadoopFetcher and related classes and tests.
* Now fetching libthrift, catalina-ant and tehuti from Maven.
Release 1.9.7 on 03/04/2015
* Improve Voldemort BuildAndPush
* Fix bugs in FailureDetector and AdminConnectionVerifier for slop streaming
* Fix log messages and English typo
* Fix NIO client thread, selector and admin thread names
Release 1.9.6 on 02/13/2015
* Bug fix for diverging metadata versions.
* IntelliJ support in gradle build.
* Avoid extra allocation on AvroGenericVersioned.
* Unit test fixes.
* Dependency upgrades:
* Jackson moved to 1.9.13
* Azkaban moved to 2.5.0, fetched from Maven Central, and removed from private-lib.
* Build and Push clean ups and improvements:
* Cleaned up config strings into constants.
* Sanity checks and error messages for bad avro configs.
* Removed proprietary LinkedIn monitoring code.
* Added open-source friendly hooks for running arbitrary code in BnP.
Release 1.9.5 on 01/12/2015
* Version metadata key fix for incorrectly incrementing on each restarts
* ReplaceNodeCLI fixes and adding post conditions.
* Tests refactoring and a new class ClientTrafficGenerator for sending traffic
* Fixed Cluster.equals to do the node state check
* Enhanced meta check-version to do the auto analysis and give output
* Readonly replication helper tool to output, to boot strap the dead node
* Windows batch file equivalents for new command functionalities
* DeleteKeysCLI tool to delete the serialized keys from Voldemort
* Krati entries iterator bug fix
Release 1.9.4 on 12/08/2014
* Add separate failure detector for slop pusher job so that metadata slop.streaming.enabled functions.
* Rename StoreVerifier to ConnectionVerifier.
* Create AdminConnectionVerifier and AdminSlopStreamingVerifier for failure detectors under OFFLINE_SERVER.
Release 1.9.3 on 11/25/2014
* Add support for disabling the client traffic (Offline server) to do maintenance on the server.
* Add support for enabling/disabling slop streaming, partition streaming, ro fetch independently.
* VAdmin support for offline server and other switches.
* ReplaceNodeCLI command to swap a node in the cluster with other voldemort node.
* Additional logging for failure detector.
Release 1.9.2 on 10/07/2014
* Fix Coordinator related tests
Release 1.9.1 on 10/03/2014
* Admin service and client for Coordinator
* Some refactoring
* Fix getmetadata shell command
* Fix origin time in coordinator rest request
Release 1.9.0 on 09/22/2014
* Updating release notes for 1.9.0
Release 1.8.16 on 09/04/2014
* QuotaException causes delete failure.
* bdb native backup fixups
Release 1.8.15 on 09/02/2014
* Fixed double-counting stat bug.
* Merged GET/GET_ALL into READ quotas and PUT/DELETE into WRITE quotas.
* Cleaned up the slop delete code.
* Fixed the C++ build so it works on more recent OS.
Release 1.8.14 on 08/26/2014
* Tehuti 0.5 that includes a fix about Histograms
Release 1.8.13 on 08/25/2014
* Fixes duplicated error message for coordinator
* Adds zone checks
* Makes tools that depend on zone 0 to be available
* Polishes Java client example
* Makes delete op use getVersion instead of get
Release 1.8.12 on 08/11/2014
* Speeds up vector clock serialization.
* Adds a clean up task in the Gradle build that fixes an intermittent issue with builds on mac.
* Expands test configurations.
* Some fixes and clean ups in the C++ client.
* Rewrites RequestCounter so that it leverages the Tehuti metrics lib.
Release 1.8.11 on 08/01/2014
* Expose http decoder parameters as config
* Client Parallel request Exceptions are not reported to failure detector
* ClientRequest Executor has infinite timeout and log cleanup
* Threshold failure detector bug fixes
Release 1.8.10 on 07/23/2014
* Registering slops for parallel puts that fail on QuotaExceededException
Release 1.8.9 on 07/02/2014
* Adding backwards compatibility checks in AdminClient
Release 1.8.8 on 06/24/2014
* Wrap mime multipart into a mime message to ensure headers are updated in
REST ResponseSenders
* Fix transport client instantiation and HttpClientFactory shutdown
* Fix build.gradle for setting java builder property in .project file
Release 1.8.6 on 06/11/2014
* Fix related to Quota metrics (store level instead of aggregate)
Release 1.8.6 on 06/11/2014
* Fix related to quotas - checking per store instead of aggregate
Release 1.8.5 on 06/06/2014
*Gradle related changes
- empty settings file
- empty voldemort-contrib directory
- gradle wrapper to fetch required version
Release 1.8.4 on 06/03/2014
* Fix RESTClientFactory for a clean shutdown
* Add check for SocketAndStreams in sendAndReceive
* Added the ability to query a single store using AdminClient / admin tool
Release 1.8.3 on 05/27/2014
* Add debug log messages for CoordinatorService
* Fix NetworkClassLoader bug when running in Windows OS (Carlos Tasada)
* Add junit support for gradle
* Fix bind failures while running tests
* Add overwrite option to fork lift tool
* Use standard logger in the StreamingClient constructor
* [python client] Fixed reconnect in _execute_request (Søren Holbech)
* Fix the admin metadata check command and did same change for old admin command.
Release 1.8.2 on 05/09/2014
* Bug fix in MetadataStore regarding updation of routing strategy
Release 1.8.1 on 05/01/2014
* Fix AbstractStoreClientFactory for backwards compatibility
* Cleaned up logging messages
* Unit tests for big Binary data
* Refactored voldemort admin tool - Phase 1 and Phase 2
Release 1.8.0 on 04/21/2014
* Split the stores.xml into individual stores on Voldemort server
* Maintaing backwards compatibility of MetadataStore
* DefaultStoreClient bootstraps using store name (instead of stores.xml)
* New Admin tool functionality to update stores (instead of replace)
Release 1.7.3 on 04/16/2014
* Fix RESTClientFactory to return a LazyStoreClient
Release 1.7.2 on 04/15/2014
* Quota limiting related fix
* Coordinator related fix
Release 1.7.1 on 04/10/2014
* Same as 1.7.0
Release 1.7.0 on 04/09/2014
* Improves BDB coversion process and documentations
* Adds scala build support and shell (Saurabh Bhatia)
* Add removeStorageEngine call to StorageConfiguration
Release 1.6.9 on 03/19/2014
* Fix a big where StreamingClient will throw NPE if not initialized
* Fix multiple broken tests
* Modified RO Rebalance Test to cover one more case
* Adding thin client shell for Voldemort Coordinator
* Move to tusk-0.0.2
Release 1.6.8 on 02/21/2014
* Add a tool to dump BDB data to Text and the reverse
* Fix racing condition problem in CachingStoreClientFactory
* Testing, client shell and admin tool enhancements from suletm,
ahimta, bhsaurabh and Holden Caufield
* ConsistencyCheckTool enhancement
* Fix bug in Coordinator for enabling different startup ports
Release 1.6.7 on 02/07/2014
* Updating the ivy file to refer to tusk-0.0.2 (build fix for nuage)
Release 1.6.6 on 02/07/2014
* Quota management in Voldemort
* Creation of read only stores in RO2 format
* Updation of query key feature in admin tool
Release 1.6.5 on 02/04/2014
* Adding AdminClient functionality to update store definitions
Release 1.6.4 on 02/03/2014
* Fix bug when RO server NPE when rolled back
* RO bug fix caused by RW optimization code
Release 1.6.3 on 01/24/2014
* Fix KeyVersionFetcher to not depend on nodeid 0
Release 1.6.2 on 01/10/2014
* Slops stored for nodes no longer in cluster, will die automatically
* Admin command to selectively purge slops destined to nodes/zones/stores
* Admin command to atomically update cluster.xml, stores.xml in one go
* Added admin command show-routing-plan to trace routing table for a key
* ZoneClipperCLI also adjusts store definitions based on dropped zone
* Numerous tests around non contiguous node ids/zone ids
Release 1.6.1 on 12/13/2013
* Avro support for fetch keys/entries admin command
* Store Clients are now cached by default, within the same factory
* Rigoruous tests around non contiguous node ids/zone ids
* ZenStoreClient efficient resource managerment
Release 1.6.0 on 11/26/2013
NOTE: This is an open source release! This release can be downloaded here:
https://github.com/voldemort/voldemort/releases
Changes made since 1.5.9:
* Fixes to remove dependency on non contiguous node ids
* Stats for store client
Release 1.5.9 on 11/05/2013
* Bug fix in AsyncMetadataVersionManager (related to store version tracking)
Release 1.5.8 on 10/29/2013
* Replace resthdfs jar with tusk.jar
* Incorporate retries into rest hdfs fetcher
Release 1.5.7 on 10/28/2013
* Voldemort now uses BDB 5.0.88
* Support for RO fetches via RestHdfs
* Fix NPE in Streaming Client
* Retrying fetches on all exceptions w/ delay
Release 1.5.6 on 10/18/2013
* Removed getStoreClientFactoryStats from StoreClientFactory interface (fix)
Release 1.5.5 on 10/18/2013
* Client shell with avro support
* Fix bind exceptions
* Fix ObsoleteVersion Exceptions
* Build once and push to multiple clusters
* Improvements to failure detector
* Fix parallel puts
* Prevent NPE in worker thread when shutdown
* Bug fixes in repair and prune job
Release 1.5.3 on 09/19/2013
* Bug fixes (NPE during shutdown
* Client side visibility changes (Histogram, Connect exception)
* Adding PruneJob and timebased resolving mechanism to streaming client
* R2Store cleanup required for the rest client
* Monitoring improvements (aggregated BDB stats, streaming stats)
* Monitoring and Config improvements for REST service
* Removed donor based rebalancing code
* Removed replica type from the code base
* Improvements to Rebalance controller
Release 1.4.7 on 09/04/13
Changes
* Added the Zone affinity feature on the client side
* Refactoring the rest package (Rest server and Coordinator)
* Using standard Netty pipeline in the Coordinator
* Added monitoring hooks for the Rest Service
Release 1.4.6 on 07/24/2013
Changes
* Cleanup of R2Store to work with the REST server
* Modified ZenStoreClient to avoid creating SystemStores during each
* re-bootstrap
* Added slop streaming functionality
Release 1.4.4 on 07/08/2013
Changes made since 1.4.3
* Change Repartioner to swap among nodes from all specified zones
* Added rebalance shuffle, cluster expansion scripts
Release 1.4.3 on 07/03/2013
Note: Server changes are not backwards compatible. To use new
rebalancing tooling, servers must be upgraded before hand.
* Rebalance features
- Proxying / abortable rebalance
- added proxy puts which rebalance safely abortable
- improved proxy gets to check locally first and only fetch remote
if not yet local
- Repartitioning: improved algorithms to optimize partition layout
- Plan: avoids cross-zone data movement in most rebalance cases. Only
expanding into a new zone requires data movement across zones.
- Controller
- better / more accurate progress monitoring.
- 'proxy pause' before rebalancing. Ensures clients pick up new
metadata and allows performance with proxy'ing to be observed
before servers start rebalancing.
- NOTE: Donor-based rebalance is currently broken by these changes.
* Rebalance tooling
- PartitionBalanceCLI: more nuanced and verbose analysis of cluster
balance
- RepartitionerCLI: Stand alone tool to determine repartitioning
- Stand alone 'repartitioning' scripts for key use cases: new cluster,
shuffle (existing cluster), cluster expansion, and zone expansion.
- RebalancePlanCLI: Stand alone plan tool to determine cost of a
specific rebalance
- RebalanceControllerCLI: Stand-alone tool for executing a specific
rebalance (plan)
* Administrivia
- java 7 compilation
- moved to Guava from Google Collections
- better unit test coverage of zoned clusters
- added example prod configuration
- added tools directory
- .gitignore improvements
* Bug fixes
- fixed concurrency bugs in metadata store
- fixed vector clock comparisons
- fixed many NPEs
- write slops in more cases when they are needed
- KeySampler/Fetcher handle larger working sets with OOME
Changes made since 1.4.2
* Minor code fixes
* Faster creation of BaseStoreRoutingPlan objects in the fast rebalancing path
Release 1.4.2 on 06/27/2013
* Fixing costly StoreRoutingPlan object construction
Release 1.4.0 on 06/21/2013
* Zone expansion release
Release 1.3.4 on 06/19/2013
* Read Write store bug fixes
- Rewrite of InMemoryStorageEngine + config to control multiVersionPuts
- Fixed a bug in the read-repair logic which was causing unnecessary puts.
- Fixed put operation in PipelineRoutedStore to ensure slop is submitted
- Fixed VectorClock bug triggered during versioned puts
* Rebalance improvement
- Made it safe to abort rebalance
- Proxy puts are established so that 'old' partition is updated and so an aborted rebalance can "roll back" without any data loss
- Proxy gets & puts are established within the zone making improving performance during rebalance
* Coordinator
- Added monitoring capability in the Coordinator
Release 1.3.3 on 04/24/2013
* VoldemortBuildandPush
- Fixed bug with schema check
* Streaming Client
- Fixed issue with redundant callback invocation
Release 1.3.1 on 03/25/2013
* HDFSFetcher
- Fixed the bug in calculating checksums when we entere a retry loop
- refactored per file checksums
- added junit test case to simulate intermittent IO exceptions
* voldemort.client.protocol.admin.AdminClient
- Added AdminStoreClient so that AdminClient can do store operations
against specific store on a specific node.
- Added helper methods for diong put & get for specific node/store
- Added voldemort.client.protocol.admin.QueryKeyResult type to
simplify QueryKey interface
* Improved FetchStreamRequestHandler and sub-classes
- Renamed all sub-classes: 'FullScan' and 'PartitionScan' prefixes
for pertinent stream request handler implementations.
- Removed unused skipRecords parameter.
- Added recordsPerPartition parameter to fetch limited portion of
each partition.
- All logic about how many keys to fetch (return to invoker) is
server side now.
* RebalanceCLI
- Added many options to help improve the balance of (zoned) clusters.
- Analysis of the balance of a cluster is significantly more detailed.
- Fixed a bug that reduced the balance of a cluster each time it was
expanded.
- Many different algorithms for improving cluster balance are
implemented in voldemort.utils.RebalanceClusterUtils
* ConsistencyCheck & ConsistencyFixCLI
- New tools for ensuring data durability. These tools are necessary
because slop creation can fail during put operations.
- ConsistencyCheck determines which keys, if any, lack
"consistency". I.e., are present on only a subset of the expected
partitions.
- ConsistencyFix takes a list of bad (inconsistent) keys and makes
sure they are present on all expected partitions.
- ConsistencyFix also has an interface for repairing "orphaned" keys
that could result from an aborted rebalance.
* KeySamplerCLI & KeyVersionFetcherCLI
- KeySamplerCLI is a new tool that reads some number of keys for
specified partitions/stores.
- KeyVersionFetcherCLI is a new tool that, given a key and a store,
fetches the version from all nodes that host a partition that
ought to store a replica of the key's value.
- Together, KeySamplerCLI and KeyVersionFetcherCLI correctly
implement the intended functionality of the Entropy tool (for
servers that implement either FullScan and PartitionScan fetches).
- Entropy tool had been used in the past to verify a sample of keys
before and after a rebalance. Entropy tool does not work as
intended/expected. This is exacerbated by the partition aware
layouts. Instead of trying to fix the Entropy tool, these two new
tools were developed. Entropy is deprecated and will eventually be
removed from the code base.
* Substantial refactoring of helper & util methods:
- voldemort.cluster.Cluster : added helper methods
- voldemort.utils.ClusterInstance : wraps up one Cluster &
List<StoreDefinition>
- voldemort.utils.Cluster : utils for single Cluster object.
- voldemort.utils.NodeUtils : utils for Node object.
- voldemort.utils.RebalanceUtils : Many methods moved to more
appropriate helper classes
- voldemort.utils.StoreDefinitionUtils : utils for StoreDefinition
object.
- voldemort.utils.StoreInstance : wraps up one Cluster & one
StoreDefinition
* Et cetera
- ByteUtils toHexString & from HexString now rely on standard
libraries
- voldemort.client.AdminFetchTest now tests FullScan and
PartitionScan fetches
- voldemort.store.routed.ReadRepairerTest annotated all tests with
@Test
Release 1.3.0 on 03/08/2013
NOTE: This is an open source release! This release can be downloaded here:
http://github.com/voldemort/voldemort/downloads.
Changes made since 1.2.3
* VoldemortConfig and ClientConfig now contain detailed documentation
* BDB-JE defaults set to ones in prod@linkedin
* Bug fixes on kerberos support for Hadoop
Release 1.2.3 on 02/20/2013
Changes made since 1.2.2
* Added a retry loop and synchronized block while getting Hadoop FS
* Code cleanup in HdfsFetcher to make it more readable.
* Throwing explicit exceptions in HdfsFetcher instead of
returning null to be more precise in the Azkaban logs.
Release 1.2.2 on 02/19/2013
Changes made since 1.2.1
* Synchronized the streaming API
* Fixed some of the streaming API tests.
Release 1.2.1 on 0/30/2013
Changes made since 1.2.0
* Added a Streaming API and related tests.
* Refactoring of the admin client apis into functional inner classes
Release 1.2.0 on 01/21/2013
Changes made since 1.1.9
* Added an Admin API to fetch orphaned key / entries
* Improved some tests related to streaming API.
* Correcting commons-codec version in ivy file (1.4)
Release 1.1.9 on 01/15/2013
Changes made since 1.1.8
* Asynchronous socket checkout improvements
* Changed checkout behavior of KeyedResourcePool to only create new
connections when there are no resources available (rather than
creating new connections until the pool is full)
* Changed QueuedKeyedResourcePool.reset behavior to better match
KeyedResourcePool (i.e., to not cancel queued asynchronous
requests unnecessarily)
* Removed (unnecessary) synchronization primitives from keyed resource pool
* Reduce granularity of failure detector locking within ThresholdFailureDetector
* Minor features/improvements
* Less verbose logging in the face of expected exceptions and errors
* Refactored (Queued)KeyedResourcePoolTest
* Bug fixes
* Fixed possible race condition for resource creation in KeyedResourcePool
* More efficient (time & space) and simpler Histogram implementation
with improved tests
Release 1.1.8 on 01/14/2013
Changes made since release 1.1.7
* Enhanced Server Monitoring
-- Server NIO layer
-- Streaming operations to the server
-- BDB storage exception counts
* Ability to turn off BDB Checkpointing during batch modifications
* Added ability to delete old checksum files in Build and Push reducer
* Upgrade Hadoop jar to 1.0.4-p2
Release 1.1.7 on 01/03/2013
NOTE: This release is based off of release 1.1.4
Changes made since release 1.1.4
* Upgrading Hadoop jar to 1.0.2
* Added support for Kerberos authentication in HdfsFetcher
* Extra config parameters for Kerberos config and keytab file
NOTE: Release 1.1.5 and 1.1.6 are special client side releases
not based off of master. 1.1.5 was rolled back to to a weird bug.
1.1.6 is a special client side release including Auto-
bootstrapper and Versioned Avro support.
Release 1.1.4 on 11/29/2012
Changes made since release 1.1.3
* Added BDB parameters to control LRU behavior in cache & proactive cleaner migration
* Added a mlock fix for pinning the indexes of RO stores in memory
Release 1.1.3 on 11/28/2012
Changes made since release 1.1.2
* Fixed a bug in the build and push job, specifically the Mapper
that caused collisions
* Added retry mechanism with the HDFS fetcher for hftp
Release 1.1.2 on 10/31/2012
Changes made since release 1.1.1
* Reverted a change to voldemort.versioning.Versioned.getVersion() so
that a Version is returned as our clients expect.
Release 1.1.1 on 10/30/2012
Changes made since release 1.1.0
* Fixed connection leak in ClientRequestExecutorFactory
* Changed client to default to DefaultStoreClient
Release 1.1.0 on 10/19/2012
Changes made since release 1.0.0
IMPORTANT NOTE : This release has significant changes to the BDB storage layer.
Users are required to read the bin/PREUPGRADE_FOR_1_1_X_README file
thoroughly before attempting to upgrade to 1.1.0. The necessary data
conversion will be done through bin/voldemort-convert-bdb.sh
* Upgrading to JE 4.1.17
* New data format that handles conflicting updates in Voldemort more
efficiently
* Move data off heap and only use it for Index
* When scanning, evict whatever you bring in right away.
* Partition based scan api to dramatically speed up rebalancing & restore
using Partition aware scans (you exactly scan whatever you want to fetch)
* Flexible knobs to control scheduling of DataCleanupJob
Release 1.0.0 on 10/17/2012
NOTE: The large version number jump from 0.96 to 1.0.0 is to
standardize on a version number of the sort MAJOR.MINOR.PATCH. This
change is part of our effort to treat internal and open source
releases in a much more similar manner. Along these lines, release
notes for internal releases (like this one) are committed on the
master branch. We hope this improves transparency as we work towards
the next open source release.
Changes made since release 0.96
* Auto bootstrapping: ZenStoreClient and System stores
* Added server side system stores for managing metadata
* ZenStoreClient interacts with system stores
* ZenStoreClient auto bootstraps whenever cluster.xml or stores.xml changes
* Added a new routing strategy to route to all with local preference
* Added a client-registry for publishing client info and config values
* Updated LazyClientStore to try to bootstrap during Init
* Modified Failure Detector to work on a shared cluster object reference
* Avro: schema evolution and read only support
* Added new Avro serializer type that supports schema evolution
* Added Avro support to read only stores
* Added LinkedIn build-and-push Azkaban jobs to build read only stores to contrib
* Added a schema backwards compatibility check to VoldemortAdminTool and on server startup to prevent mishaps due to bad schemas
* Non-blocking IO: Fixed design flaw that blocked in the face of slow servers
* Asynchronous operations no longer do a blocking checkout to get a SocketDestination
* Added additional stats collection for better visibility into request queues
* Minor features
* Enhanced VoldemortAdminTool to update store metadata version
* Enhanced VoldemortAdminTool to work with the new system stores
* Added feature to voldemort-shell.sh to dump byte & object arrays
* Added a SlowStorageEngine for testing degraded mode performance
* Added mechanism to isolate BDB cache usage among stores
* Enhanced debug logging (for traffic analysis).
* Python client bug fixes (from pull request)
* Improved messages in request tracing
* Cleaned up help/usage messages within the client shell
* Added server config to control socket backlog
* Added "--query-keys" option to query multiple keys of multiple stores from specific node
* Added control to DataCleanupJob Frequency
* Unified jmxid as the factory across the board
* Tools
* bin/generate_cluster_xml.py to generate cluster.xml
* bin/repeat-junit.sh and bin/repeat-junit-test.sh to repeatedly run tests
* Bug fixes
* Changed getall return behavior to comply with javadoc
* Fixed a bug that caused unnecessary serial requests in getall
* HFTP performance issue bug fix (fix in byte buffer and copy process)
* Fixed a bug that prevented "--fetch-keys" and "--fetch-entries" in admin tool from showing multiple store results
* Fixed problem in sample config that prevented the server from starting
* Fixed some intermittent BindException failures across many unit tests
* Fixed some intermittent rebalance test failures
* Wrapped long running tests with timeouts
Release 0.96 on 09/05/2012
Changes made since 0.90.1
* Monitoring:
* Append cluster name to various mbeans for better stats display
* Implement average throughput in bytes
* Add BDB JE stats
* Add 95th and 99th latency tracking
* Add stats for ClientRequestExecutorPool
* Add error/exception count and max getall count
* BDB+ Data cleanup Monitoring changes
* Rebalancing:
* Donor-based rebalancing and post cleanup (see https://github.com/voldemort/voldemort/wiki/Voldemort-Donor-Based-Rebalancing for more details)
* Rebalancing integration testing framework (under test/integration/voldemort/rebalance/)
* Generate multiple cluster.xml files based on the number specified when running the tool and choose the cluster with the smallest std dev as the final-cluster.xml
* Add status output to log for updateEntries (used by rebalancing)
* Read-only pipeline:
* Add hftp and webhdfs support
* Read-only bandwidth dynamic throttler
* Add minimum throttle limit per store
* Add rollback capability to the Admin tool
* Voldemort-backed stack and index linked list impl
* Change client requests to not process responses after timeout
* Modified client request executor timeout to not factor in the NIO selector timeout
* Added BDB native backup capabalities, checksum verification and incremental backups (well tested, but not yet used in production)
* Add additional client-side tracing for debugging and consistency analytics
* Clean up logging during exception at client-side
* Security exception handling
* Add snappy to CompressionStrategyFactory
* Add configurable option to interrupt service being unscheduled
* Add logging support for tracking ScanPermit owners (for debugging purposes)
* Add a jmx terminate operation for async jobs
* Add zone option for restore from replicas
* Changing the enable.nio.connector to true by default
* Better disconnection handling for python client
* Split junit tests into a long and a short test suites
* Add separate timeouts for different operations (put, get, delete, and getAll
* Allow getAll to return partial results upon timeout
* Improved cluster generation tool
* Added log4j properties folder for junit test
* Bug fixes:
* httpclient 3.x to httpclient 4.x
* Fix NPE in listing read-only store versions
* Fixed 2 failure detector bugs during rebalancing or node swapping
* Fixed a thread leak issue in StreamingSlopPusher
* Fixed a NIO bug
* Fixed a bug in TimeBasedInconsistency resolver.
* Fixed race condition in client socket close
* Fixed a potential deadlock issue in ScanPermitWrapper
* Fixed a bug where a read returns null (on rare occations) when being concurrent with a write
* Fixed a performance bug in HdfsFetcher when hftp is used
Changes made since 0.90
* Updated the documentation for Voldemort shell tool in NOTES
* Added Admin API to perform Bdb data cleanup (repairJob)
and corresponding unit tests
* Fixes in restore from replication, store creation code
* Improved failure detector configuration. ThresholdFailureDetector
is now the default option
* Multiple Fixes to the Voldemort ruby client
* Added additional Jmx metrics to expose Bdb environment statistics,
caching statistics and Voldemort batch operation statistics
* Updated default timeout for restore 'from replica' to 365 days
* New feature in the performance tool: '--use-sample' option enables
'read, write back unmodified' transactions in place of writes in
the workload (allows for testing read-write transactions on stores
with complex schemas)
* Added the ability to dynamically update cluster.xml and
reinitialize the scheduler
Release 0.90 on 7/10/2011
Changes made since 0.81
* All upgrading instructions can be found here - https://github.com/voldemort/voldemort/wiki/Upgrading-from-0.81
* Tooling
- Single consolidated administrative tool - Got rid of the admin client
shell and have a better Voldemort admin tool. More documentation -
https://github.com/voldemort/voldemort/wiki/Voldemort-Admin-tool
* Web manager
- Basic GUI in JRuby ( and Sinatra ), capable of querying for store metadata.
- Read contrib/web-manager/README.md for more details
* Rebalancing
- New better rebalancing support with (a) checkpointing (b) progress bar (c) support for RO store rebalancing
- Documentation - https://github.com/voldemort/voldemort/wiki/Voldemort-Rebalancing
- In the process solves Issue 203, 288, 305, 307, 311
* Client side changes
- Pipeline routed store ( with support for topology awareness ) - Changed the default client side routing
to use a state machine like system. More documentation - https://github.com/voldemort/voldemort/wiki/RoutedStore-redesign
- Lazy store client - Some of our clients are very inactive and don't really do many requests. For such
clients it doesn't make sense to bootstrap the metadata at startup. Hence from now onwards clients instantiated
from SocketStoreClientFactory give a LazyStoreClient which will not bootstrap till the first request is made
- Caching store client factory - We have also checked-in a new store client factory which we use internally
at LinkedIn to cache store clients for stores which we repeatedly.
- Failure detector - We have fixed some bugs in the ThresholdFailureDetector. This is an improvement over the naive
BannageFailureDetector which would aggressively mark nodes down after a single failure. More details -
https://github.com/voldemort/voldemort/wiki/Client-side-failure-detector-implementations
- Issue 219: StoreClient::put returns a Version - You may have to upgrade your clients since now the StoreClient returns
a version to be used in successive requests.
* Read-only store changes
- Admin based store swapper - Till 0.81 we relied on a servlet based fetcher / swapper.
Now we support an admin based store swapper which reports progress.
- Other changes - (a) New directory format (b) Two new data format. Support for iterating over read-only data
(c) Checksum of data in .metadata file (d) Some monitoring changes of RO stores
* Jar upgrades
- Introduction of JNA 3.2.7 - We have introduced JNA 3.2.7 in this release for supporting symbolic links in read-only stores.
More details about the new RO format and changes can be found here - https://github.com/voldemort/voldemort/wiki/Upgrading-from-0.81
- Protocol Buffers 2.3.0 - Upgrading protocol buffers from 2.2.0 to 2.3.0
- Avro 1.4.0 - Upgraded Avro from 1.3.0 to 1.4.0
* Storage engine
- Added support for Krati 0.3.6 as a storage engine ( http://sna-projects.com/krati/ ) -
https://github.com/voldemort/voldemort/tree/release-090/contrib/krati
* Core features
- Hinted handoff - https://github.com/voldemort/voldemort/wiki/Hinted-Handoff
- Experimental support for server side transforms - https://github.com/voldemort/voldemort/wiki/Server-side-transforms-in-Voldemort
- Topology awareness ( i.e. datacenter / rack ) - https://github.com/voldemort/voldemort/wiki/Topology-awareness-capability
- Repair Job ( Job which will delete data if the node is not responsible for it )
* Better monitoring
- Tons of JMX changes ( average bytes, etc ) - More details ( https://github.com/voldemort/voldemort/wiki/JMX-Monitoring )
- Key distribution generator - Ability to estimate skew of your cluster ( ./bin/run-class.sh voldemort.utils.KeyDistributionGenerator )
* Clients
- Python - Updated Python client with support for Binary JSON serialized stores ( ./clients/python/README )
- Ruby - Updated Ruby client ( ./clients/ruby/README.md )
Release 0.81 on 6/15/2010
Changes made since 0.80.2:
* IMPORTANT: we have upgraded the Hadoop Core jar to v0.20.2. Since
this version of Hadoop requires Java 6, in order to retain backwards
compatibility with Java 5, you will need to replace this jar with
the Hadoop 0.18.* core jar.
* Multiple donors for Voldemort Rebalancing: speeds up rebalancing, by
allowing a single stealer node to transfer partitions from multiple
nodes
* Features for EC2 Testing
* Read-only Stores: backwards compatibility with 0.70, bug fix in
Checksum code
* Hadoop InputFormat, Pig LoadFunc in Contrib: ability to perform
Map/Reduce (either directly via Hadoop or using Pig) over data in
Voldemort stores. See:
< contrib/hadoop/test/voldemort/hadoop/VoldemortWordCount.java > for
an example.
* Voldemort Performance Tool: performance measurement tool based on
YCSB (Yahoo Cloud Serving Benchmark) code. Run
< bin/voldemort-performance-tool.sh --help > for more information.
* Reverted default failure detector implementation back to
BannagedPeriodFailureDetector due to potential bugs in the
ThresholdFailureDetector
Release 0.80.2 on 4/27/2010
Changes made since 0.80.1:
* Batched NIO writes (improves performance for AdminClient/Streaming
operations when the NIO connector is enabled)
* VoldemortAdminTool: Added support to specify stores, support to
fetch all keys to a binary file, support to save keys in ASCII
(JSON) format [experimental], capability to add stores. Run
< bin/voldemort-admin-tool.sh --help > for more information.
* Minor bug fixes for Rebalancing
* Issue 240: Voldemort fetcher should use different temp directories for
different stores
* Checksum capability during construction of Voldemort read-only stores
Release 0.80.1 on 3/23/2010
Changes made since 0.80:
* Issue 133: Support for Apache Avro as a serialization format
* Issue 223: Changed the default client to use
TresholdFailureDetector
* Fixed issue 222: Revised KeyedResourcePool.close(K key) to fix
leaking sockets
* Fixed issue 198: Revised ReadRepairer to use a separate copy of the
vector clock, fixing a situation where NoSuchElementException would
be thrown
* Miscellaneous enhancement: support for TCP keep-alives, improved
read only store utilities, command line interface to AdminClient,
improved load testing tools
Release 0.80 on 2/18/2010
Changes made since 0.70.1:
* IMPORTANT: backwards compatibility between the client and server has
changed. A backwards incompatibility in the wire protocol was found
between releases 0.60 to 0.70.1 and releases prior to 0.60. We chose
to make 0.80 compatible with 0.57.1 and earlier versions, while
introducing an incompatibility with 0.60-0.70.1. What this means is
that if you're presently running 0.60 and higher, you would need to
upgrade the Voldemort jar files on *all* servers and clients.
* Upgraded the BDB storage engine to use BerkeleyDB-JE 4.0.92,
retaining ability to use BerkeleyDB-JE 3.3.* if desired. IMPORTANT:
if one switches from BerkeleyDB-JE 3.3.* to 4.0.92 they will be able
to access all of existing data. Once a switch has been made to
4.0.92 the data will not be readable by earlier versions of BDB. If
there's a chance that a roll back to 3.3.* might be needed, the best
course of action will be to make a backup of existing data
prior to upgrading.
Switching between 3.3.* and 4.0.* would also require rebuilding the
class files (e.g., by running "ant clean && ant release" after
replacing the BDB jar files).
* Compression support for read-only stores
* Increased the socket buffer size for transferring read-only
stores from Hadoop for improved performance over high-latency links
* NIO support for the Admin Service, including Streaming
Functionality
* Support for adding stores on the fly via the Admin
Service
* Fixed issue 209: Incorrect object passed to List.contains in
RebalanceUtils.getLatestCluster()
* Fixed issue 211: Unnecessary read repairs during getAll() with more
than one key
* Other enhancements: better CLI for rebalancing, throttling in
Admin Service is now based on all disk activity
Release 0.70.1 on 2/1/2010
Changes made since 0.70:
* Fixed issue 205: if no keys passed to getAll() were in partitions
undergoing rebalancing, proxyGetAll() would be called with an
empty list even if rebalancing wasn't happening
Release 0.70 on 1/27/2010
Changes made since 0.60.1:
* A beta of rebalancing (dynamic cluster expansion) support merged
into the main branch. See the project's wiki for more information:
http://wiki.github.com/voldemort/voldemort/voldemort-rebalancing
* New failure detector merged into the main branch:
http://wiki.github.com/voldemort/voldemort/failure-detection
* Beta mechanism for restoring all of node's data from replicas on
demand. This is an alternative to a more gradual mechanism provided
by read-repair: useful when a machine is down for a prolonged period
and is then re-inserted into the cluster.
Invoked via JMX: the operation is restoreDataFromReplication in the
voldemort.server.VoldemortServer MBean, with a mandatory parameter
(integer >= 1) indicating the number of transfers to do in parallel.
* Simple gossip protocol (for cluster metadata) merged
into the main branch. Disabled by default: use "enable.gossip=true"
to enable, use "gossip.interval.ms" to set an interval at which gossip
occurs (default: 30000 i.e., 30 seconds).
* Fixed issue 190: add a way of aggregating performance data over
all stores
* Fixed issue 181: stack trackes shouldn't be filled for Obsolete
version exception
Release 0.60.1 on 12/18/2009
Minor changes made since 0.60:
* Better logging in the exception thrown if config/.temp and
config/.version are copied
* Bumping up the version to 0.60.1 in order to release updated
archives, fixing an error in the stores.xml for single_node_cluster
sample config
Release 0.60 on 12/15/2009
Changes made since 0.57.1:
* Admin Client/Server API: adds support for streaming-based transfer
of entries between nodes, deleting entries on remote nodes,
remotely deleting and updating metadata
* EC2 testing: a way to periodically run integration and performance
tests which involve Voldemort instances on different machines
* Experimental support for views
* Interpolation search for read-only stores
* Support for large lists and strings in the JSON serializer
* LZF compression support
* Ruby client contributed
* Fixed issue 170: hanging if a port is used by another process
* Fixed issue 122: suspicious integer division in
RequestCounter.getThroughput
* Miscellaneous improvements and bug fixes for read-only stores
* Fixed issue 168: added StorageEngine.keys()
Release 0.57.1 on 11/27/2009
Minor change made since 0.57:
* Modified build.xml to exclude .git directory from release tarballs/zipfiles
Release 0.57 on 11/16/2009
The following changes were made since 0.56:
* Fixed an issue in ReadOnlyEngine's close() method
* Fixed hidden logging in StorageService
* Fix for issue 163 (lock mode during get)
* Exposed bdb environment stats with setFast(false)
Release 0.56 on 10/26/2009
The following changes were made since 0.55:
* Fix for issue 164: Changed default bdb.max.logfile.size to 60MB
* Make file deletes asynchronous in read only store swap
* Added better debug logs for bdb stats
* Fixed race condition in AbstractSocketPoolTest
* Added improved monitoring for bdb stats
* Added backoff and retry logic in bootstrap code
* Not logging obosleteVersionException(s) or counting it in JMX exception count
* Fixed issue 159
* C++ client building on OS X
* Rely only on the number of versions returned to decide whether to retrieve
the value for put(K,V)
* Implemented issue 152: getVersion() API in Store
Release 0.55 on 10/7/2009
The following changes were made since 0.52 (in summary):
* Add an event throttler
* Added DataCleanupJob
* Protocol buffers at version 2.2.0
* BDB JE upgraded to 3.3.87
* Added data compression support (enable by adding
<compression>
<type>gzip</type>
</compression>
to value-serializer or key-serializer sections in stores.xml)
* Added a resource pool/socket pool implementation (no longer using
commons-pool)
* Added server-side NIO support (enabled by setting
enable.nio.connector=true
in server.properties)
* Improved the efficiency of the protocol buffers network protocol by using
CodedOutputStream/CodedInputStream (zero copy transfers)
* Fix for issue #21: incorrectness in vector clock inconsistency resolver
* Upgrade google-collections
* Support for building read only stores in Hadoop
* Added a C++ client