00-RELEASENOTES

Redis 3.0 release notes
=======================

--------------------------------------------------------------------------------
Upgrade urgency levels:

LOW:      No need to upgrade unless there are new features you want to use.
MODERATE: Program an upgrade of the server, but it's not urgent.
HIGH:     There is a critical bug that may affect a subset of users. Upgrade!
CRITICAL: There is a critical bug affecting MOST USERS. Upgrade ASAP.
--------------------------------------------------------------------------------

--[ Redis 3.0.5 ] Release date: 15 Oct 2015

Upgrade urgency: MODERATE, the most important thing is a fix in the replication
                 code that may make the slave hanging forever if the master
                 remains with an open socket even if it is no longer able to
                 reply.

* [FIX] MOVE now moves the TTL as well. A bug lasting forever... finally
        fixed thanks to Andy Grunwald that reported it.
        (reported by Andy Grunwald, fixed by Salvatore Sanfilippo)
* [FIX] Fix a false positive in HSTRLEN test.
* [FIX] Fix a bug in redis-cli --pipe mode that was not able to read back
        replies from the server incrementally. Now a mass import will use
        a lot less memory, and you can use --pipe to do incremental streaming.
        (reported by Twitter user @fsaintjacques, fixed by Salvatore
        Sanfilippo)
* [FIX] Slave detection of master timeout. (fixed by Kevin McGehee, refactoring
        and regression test by Salvatore Sanfilippo)

* [NEW] Cluster: redis-trib fix can fix an additional case for opens lots.
        (Salvatore Sanfilippo)
* [NEW] Cluster: redis-trib import support for --copy and --replace options
        (David Thomson)

--[ Redis 3.0.4 ] Release date: 8 Sep 2015

Upgrade urgency: HIGH for Redis and Sentinel. However note that in order to
                 fix certain replication bugs, the replication internals were
                 modified in a very heavy way. So while this release is
                 conceptually saner, it may contain regressions. For this
                 reason, before the release, QA activities were performed by
                 me (antirez) and Redis Labs and no evident bug was found.

* [FIX] A number of bugs related to replication PSYNC and the (yet experimental)
        diskless replication feature were fixed. The bugs could lead to
        inconsistency between masters and slaves. (Salvatore Sanfilippo, Oran
        Agra fixed the issue found by Yuval Inbar)
* [FIX] A replication bug in the context of PSYNC partial resynchonization was
        found and fixed. This bug happens even when diskless replication is off
        in the case different slaves connect at different times while the master
        is creating an RDB file, and later a partial resynchronization is
        attempted by a slave that connected not as the first one. (Salvatore
        Sanfilippo, Oran Agra)
* [FIX] Chained replication and PSYNC interactions leading to potential stale
        chained slaves data set, see issue #2694. (Salvatore Sanfilippo fixed
        an issue reported by "GeorgeBJ" user at Github)
* [FIX] redis-cli --scan iteration fixed when returned cursor overflows
        32 bit signed integer. (Ofir Luzon, Yuval Inbar)
* [FIX] Sentinel: fixed a bug during the master switch process, where for a
        failed conditional check, the new configuration is rewritten, during
        a small window of time, in a corrupted way where the master is
        also reported to be one of the slaves. This bug is rare to trigger
        but apparently it happens in the wild, and the effect is to see
        a replication loop where the master will try to replicate with itself.
        A detailed explanation of the bug and its effects can be found in
        the commit message here: https://github.com/antirez/redis/commit/c20218eb5770b2cafb12bc7092313b8358fedc0a.
        The bug was found by Jan-Erik Rediger using a static analyzer and
        fixed by Salvatore Sanfilippo.
* [FIX] Sentinel lack of arity checks for certain commands.
        (Rogerio Goncalves, Salvatore Sanfilippo)

* [NEW] Replication internals rewritten in order to be more resistant to bugs.
        The replication handshake in the slave side was rewritten as a non
        blocking state machine. (Salvatore Sanfilippo, Oran Agra)
* [NEW] New "replication capabilities" feature introduced in order to signal
        from the master to the slave what are the features supported, so that
        the master can choose the kind of replication to start (diskless or
        not) when master and slave are of different versions. (Oran Agra,
        Salvatore Sanfilippo)
* [NEW] Log clients details when SLAVEOF command is received. (Salvatore
        Sanfilippo with inputs from Nick Craver and Marc Gravell).

--[ Redis 3.0.3 ] Release date: 17 Jul 2015

Upgrade urgency: LOW for Redis and Sentinel.

* [FIX] Fix blocking operations timeout precision when HZ is at its default
        value (not increased) and there are thousands of clients connected
        at the same time. This bug affected Sidekiq users that experienced
        a very long delay for BLPOP and similar commands to return for
        timeout. Check commit b029ff1 for more info. (Salvatore Sanfilippo)
* [FIX] MIGRATE "creating socket: Invalid argument" error fix. Check
        issues #2609 and #2612 for more info. (Salvatore Sanfilippo)
* [FIX] Be able to connect to the master even when the slave is bound to
        just the loopback interface and has no valid public address in the
        network the master is reacahble. (Salvatore Sanfilippo)
* [FIX] ZADD with options encoding promotion fixed. (linfangrong)
* [FIX] Reset aof_delayed_fsync on CONFIG RESETSTATS. (Tom Kiemes)
* [FIX] PFCOUNT key parsing in cluster fixed. (MOON_CLJ)
* [FIX] Fix Solaris compilation of Redis 3.0. (Jan-Erik Rediger)

* [NEW] Variadic EXISTS command. Now the command accepts multiple arguments
        and returns the total count of existing keys.

--[ Redis 3.0.2 ] Release date: 4 Jun 2015

Upgrade urgency: HIGH for Redis because of a security issue.
                 LOW for Sentinel.

* [FIX] Critical security issue fix by Ben Murphy: http://t.co/LpGTyZmfS7
* [FIX] SMOVE reply fixed when src and dst keys are the same. (Glenn Nethercutt)
* [FIX] Lua cmsgpack lib updated to support str8 type. (Sebastian Waisbrot)

* [NEW] ZADD support for options: NX, XX, CH. See new doc at redis.io.
        (Salvatore Sanfilippo)
* [NEW] Senitnel: CKQUORUM and FLUSHCONFIG commands back ported.
        (Salvatore Sanfilippo and Bill Anderson)

--[ Redis 3.0.1 ] Release date: 5 May 2015

Upgrade urgency: LOW for Redis and Cluster, MODERATE for Sentinel.

* [FIX] Sentinel memory leak due to hiredis fixed. (Salvatore Sanfilippo)
* [FIX] Sentinel memory leak on duplicated instance. (Charsyam)
* [FIX] Redis crash on Lua reaching output buffer limits. (Yossi Gottlieb)
* [FIX] Sentinel flushes config on +slave events. (Bill Anderson)

--[ Redis 3.0.0 ] Release date: 1 Apr 2015

>> What's new in Redis 3.0 compared to Redis 2.8?

* Redis Cluster: a distributed implementation of a subset of Redis.
* New "embedded string" object encoding resulting in less cache
  misses. Big speed gain under certain work loads.
* AOF child -> parent final data transmission to minimize latency due
  to "last write" during AOF rewrites.
* Much improved LRU approximation algorithm for keys eviction.
* WAIT command to block waiting for a write to be transmitted to
  the specified number of slaves.
* MIGRATE connection caching. Much faster keys migraitons.
* MIGRATE new options COPY and REPLACE.
* CLIENT PAUSE command: stop processing client requests for a
  specified amount of time.
* BITCOUNT performance improvements.
* CONFIG SET accepts memory values in different units (for example
  you can use "CONFIG SET maxmemory 1gb").
* Redis log format slightly changed reporting in each line the role of the
  instance (master/slave) or if it's a saving child log.
* INCR performance improvements.

>> Refactoring changes (no new features nor bug fixes)

* Blocking operations full refactoring (blocked.c)
* Client output buffer memory tracking refactored.

Changes between RC6 and 3.0.0 stable:

>> General changes

* Fixes to diskless replication. (Oran Agra)
* Test for BLPOP replication on role change. (Salvatore Sanfilippo)
* prepareClientToWrite() error handling improvements. (Salvatore Sanfilippo)
* Remove dict.c no longer used function. (Salvatore Sanfilippo)

>> Cluster changes

None

>> Sentinel changes

None

--[ Redis 3.0.0 RC6 (version 2.9.106) ] Release date: 24 mar 2015

Upgrade urgency: HIGH because of bugs related to Redis Custer and replication.

This is the 6th release candidate of Redis 3.0.0. This release fixes important
issues discovered during stress testing, and implements safest behavior
for blocking operations during clients reshardings, and a new much needed
functionality of Redis Cluster manual failovers.

In order to fix certain bugs quite a bit of refactoring was needed which
is usually non advisabble in a Release Candidate, but needed in order to
end with a clean fix.

>> General changes

* [FIX] Redis (non clustered & clustered) replication bug involving blocking
        operations: see issue #2473. (Salvatore Sanfilippo)

>> Cluster changes

* [FIX] clientsArePaused() fix crashing the old master during manual failover.
        (Salvatore Sanfilippo)
* [FIX] Lua scripts replication in Redis Cluster was totally broken.
        (Salvatore Sanfilippo)
* [FIX] Redirect clients blocked into list operations when the hash slot
        they are blocked into is migrated to another instance or the cluster
        state turns into "fail". (Salvatore Sanfilippo)

* [NEW] TAKEOVER option for CLUSTER FAILOVER implemented. It is now possible
        to fix a cluster manually in the minority side of the partition, for
        example in order to allow for multi DC setups & recovery.
        (Salvatore Sanfilippo)

>> Sentinel changes

No changes in Sentinel.

--[ Redis 3.0.0 RC5 (version 2.9.105) ] Release date: 20 mar 2015

Upgrade urgency: Moderate for Redis Cluster users, low otherwise.

This is the 5th release candidate of Redis 3.0.0, released in order to fix
a moderate bug in Redis Cluster. This RC does not shift in the future the
Redis 3.0.0 final release which is scheduled in a few days (we are in the
process of finishing the documentation for Redis Cluster).

>> General changes

* [FIX] Fix LATENCY command crash. (Salvatore Sanfilippo, thx to Ingmar)
* [FIX] Config: missing activerehashing option support in CONFIG SET added.
        (Salvatore Sanfilippo, thx to Bill Anderson)
* [FIX] Fix for backtrace generation issue. (Mariano Pérez Rodríguez, Matt Stancliff, Salvatore Sanfilippo)

* [NEW] Redis-cli --latency-dist backported from unstable.
        (Salvatore Sanfilippo)

>> Cluster changes

* [FIX] Avoid redundant SELECT in MIGRATE. (Tommy Wang, Salvatore Sanfilippo)
* [FIX] More robust slave check in CLUSTER REPLICATE. (Salvatore Sanfilippo)
* [FIX] Fixed possible Redis Cluster node crash due to wrong separation of
        concerns between getNodeByQuery() and Cluster global state update
        fnuction. (Salvatore Sanfilippo, thx to Ingmar)

* [NEW] Add command CLUSTER MYID to easily featch instance ID. (Michel Martens)

>> Sentinel changes

* [NEW] Support for CLIENT command added. It was missing in the command table.
        (Leandro López)

--[ Redis 3.0.0 RC4 (version 2.9.104) ] Release date: 13 feb 2015

Upgrade urgency: High for Redis if you use LRU eviction, low otherwise.

This is the 4th release candidate of Redis 3.0.0, it fixes problems with
LRU eviction that are not present in older release (2.8.x is not affected)
and adds new tools to inspect latency and load-test LRU.

>> General changes

* [FIX] redis-cli CSV output NIL spurious newline removed. (Matt Collier)
* [FIX] Memory efficiency test in unit test is now much faster: it affacted
        the total "make test" execution time in a bad way.  (Salvatore
        Sanfilippo)
* [FIX] Fixes and improvements to dict.c and LRU eviction. Redis 3.0.0 new
        LRU eviction had bugs creating high latency spikes when LRU was
        happening during the keys dictionary rehashing. This bug is not
        present into 2.8, was 3.0 specific. As a side effect of this issue
        dict.c is now improved, and LRU algorithm is more precise (better
        approximates true LRU). This was a joint effort, see issue
        #2306 for details. (Oran Agra, Sun He, Salvatore Sanfilippo).
        Thanks to Charsyam for spotting an integer overflow.

* [NEW] New latency tool: redis-cli --latency-dist is able to show an
        xterm-256 based spectrum of latencies over time. (Salvatore Sanfilippo)
* [NEW] redis-cli --lru-test implemented (cache workload simulator). (Salvatore
        Sanfilippo)
* [NEW] redis-cli --stat now shows LOAD when Redis is loading data.
* [NEW] Support "1G" etc. units in CONFIG SET. (Chris Lamb, Salvatore
        Sanfilippo)

>> Cluster changes

* None.

>> Sentinel changes

* None.

--[ Redis 3.0.0 RC3 (version 2.9.103) ] Release date: 30 jan 2015

Upgrade urgency: High for Redis Cluster users, low otherwise.

This is the third release candidate for Redis 3.0.0, the new RC fixes
several critical issues with Redis Cluster.

>> General changes

* [FIX] AOF bug unlikely to happen in practice and mostly harmless: child
        process segfaults when parent is not reachable via pipe. (Sun He)
* [FIX] Scripting engine now reports an error when misused with Lua debug
        hooks, instead of crashing. (Salvatore Sanfilippo)

>> Cluster changes

* [FIX] Several issues with Redis Cluster internal nodes objects handling.
        (Matt Stancliff & Salvatore Sanfilippo)
* [FIX] Improvements in the Cluster test.
        (Matt Stancliff & Salvatore Sanfilippo).
* [FIX] Cluster memory leaks / double frees (Matt Stancliff).
* [FIX] /dev/urandom surrogate for generation of unique IDs in a more
        cheap way. (Salvatore Sanfilippo)
* [FIX] Fixes and improvements to PING / PONG packets gossip sections
        in order to improve (and fix) failure detection and speedup
        cluster info propagation. (Salvatore Sanfilippo)

* [NEW] CLUSTER count-failure-reports command added. (Salvatore Sanfilippo)

>> Sentinel changes

No changes for Sentinel.

--[ Redis 3.0.0 RC2 (version 2.9.102) ] Release date: 13 jan 2015

Upgrade urgency: LOW.

This is the second release candidate of Redis 3.0.0. The major changes
are back porting of things implemented into the unstable branch while
this was still possible (with the new development model adopted only
bug fixes will be merged in the future).

RC2 also fixes a few Redis Cluster non critical bugs.

>> General changes

* [FIX] A number of minor bug fixes.

* [NEW] Diskless replication backportede.
* [NEW] Lua bitops and updated cmsgpack backported.
* [NEW] Transparent Huge Pages warnings and reporting backported.

>> Cluster changes

* [FIX] Fix PUBLISH cluster bus message count field.
* [FIX] It is no longer possible to write outside node hash slots using Lua.
* [FIX] Valgrind warnings (no actual bugs).
* [FIX] Less strict in acceptiong myself->ip if it's not populated.

* [NEW] Better testing of Lua scripts.

>> Sentinel changes

No changes to Sentinel.

--[ Redis 3.0.0 RC1 (version 2.9.101) ] Release date: 9 oct 2014

This is the first release candidate of Redis 3.0.0.

>> General changes

* [FIX] An very large number of small fixes, old and new, merged in the
        context of a the issue #1906. Please see the issue page here
        for exact credits: https://github.com/antirez/redis/pull/1906
        of each commit. (Matt Stancliff and many others).
* [FIX] SAVE is no longer propagated to AOF / slaves.
* [FIX] GETRANGE test no longer fails for 32 bit builds (Matt Stancliff).
* [FIX] Limit SCAN latency when the hash table is in an odd state (very few
        populted buckets because rehashing is in progress). (Xiaost and
        Salvatore Sanfilippo)

* [NEW] Redis is now able to load truncated AOF files without requiring a
        redis-check-aof utility run. The default now is to load truncated
        (but apparently not corrupted) AOFs, you can change this in redis.conf.
        (Salvatore Sanfilippo).
* [NEW] DEBUG POPULATE two args form implemented. It is now possible to
        call it with DEBUG POPULATE <count> <prefix>. Default prefix
        is "key:" as usually.
* [NEW] INCR: Modify incremented object in-place when possible. This results
        in speed improvements + possibly better memory locality.

>> Cluster changes

* [FIX] Cluster: claim ping_sent time even if we can't connect.
* [FIX] redis-trib should not abort easily on connection issues.
* [FIX] Cluster test: less console-spammy resharding test.
* [FIX] Fix logic to detect we are among a minority.
* [FIX] Process gossip section only for known nodes.

* [NEW] Redis Cluster is stable and tested enough, there is a clear MVP,
        so it was promoted from beta to stable.
* [NEW] New unit 09, Pub/Sub across the cluster.
* [NEW] New unit 08, update messages.
* [NEW] New cluster option to work with partial slots coverage.
* [NEW] More chatty cluster slaves when failover is stalled. They log reason
        with rate limiting, only when reason changes or a given time
        has elapsed.

>> Sentinel changes

* [FIX] Sentinel critical bug fixed: the absolute majority was computed in a
        wrong way because of a programming error. Now the implementation does
        what the specification says and the majority to authorize a failover
        (that should not be confused with the ODOWN quorum) is the majority of
        *all* the Sentinels ever seen for a given master, regardless of their
        current state.
* [FIX] Resolved a memory leak in the hiredis library causing a memory leak
        in Redis Sentinel when a monitored instance or another Sentinel is
        unavailable. Every reconnection attempt will leak a small amount of
        memory, but in the long run the process can reach a considerable size.

* [NEW] Sentinel: ability to announce itself with an arbitrary IP/port to work
        in the context of natted networks. However this is probably still
        not enough since there is no equivalent mechanism for slaves listed
        in the master INFO output. (Dara Kong and Salvatore Sanfilippo)

--[ Redis 3.0.0 Beta 8 (version 2.9.57) ] Release date: 29 jul 2014

This is the 8th beta of Redis 3.0.0.

>> General changes

* [FIX] Solaris compilation issues. (Matt Stancliff, Salvatore Sanfilippo)
* [FIX] Allow shared integer objects if maxmemory policy is not LRU based.
        (Salvatore Sanfilippo)
* [FIX] PFSELFTEST: less false positives. (Salvatore Sanfilippo)
* [FIX] Fail SYNC if background save child aborted due to a signal. (Yossi
        Gottlieb)

* [NEW] Latency framework backported from unstable branch. (Salvatore
        Sanfilippo)
* [NEW] AOF rewrite improved for latency. (Salvatore Sanfilippo)
* [NEW] Pub/Sub PING. (Salvatore Sanfilippo)
* [NEW] Much faster ZUNIONSTORE. (Kyle Hubert, Salvatore Sanfilippo)
* [NEW] Faster ll2string() implementation. (Salvatore Sanfilippo)

>> Cluster changes

* [FIX] CLUSTER RESET: Flush slave dataset on reset. (Salvatore Sanfilippo)
* [FIX] Replica migration: don't migrate to masters that never had slaves
        in the past, but only to masters that remained orphaned after
        failure events. (Salvatore Sanfilippo)

* [NEW] redis-trib: allow to reshard in non-interactive way. (Salvatore
        Sanfilippo)
* [NEW] Cluster test: unit 04, check consistency during resharding. (Salvatore
        Sanfilippo)
* [NEW] Cluster test: unit 05, slave selection. (Salvatore Sanfilippo)
* [NEW] Cluster test: unit 06, slaves with stale data can't failover. (Salvatore
        Sanfilippo)
* [NEW] Cluster test: unit 07, replicas migration. (Salvatore Sanfilippo)

>> Sentinel changes

* No Sentinel changes in this release.

--[ Redis 3.0.0 Beta 7 (version 2.9.56) ] Release date: 30 jun 2014

This is the 7th beta of Redis 3.0.0.

>> General changes

* [FIX] Scripting fixes backported from unstable, see Redis 2.8.12 changelog
        for more info. (Salvatore Sanfilippo)
* [FIX] Cancel SHUTDOWN if initial AOF is being written. (Matt Stancliff)

* [NEW] New command: COMMAND, for commands introspection (Matt Stancliff &
        Salvatore Sanfilippo)
* [NEW] hiredis: Update to latest version. (Matt Stancliff)
* [NEW] Jemalloc updated to 3.6.0. (Salvatore Sanfilippo)

>> Cluster changes

* [FIX] Cluster: clear NOADDR flag when updating node address.
        (Salvatore Sanfilippo)

* [NEW] New CLUSTER SLOTS command to simplify Cluster clients operations.
        (Matt Stancliff)
* [NEW] More Cluster tests. (Salvatore Sanfilippo)
* [NEW] Log when failover authorization are granted / denied.
        (Salvatore Sanfilippo)

>> Sentinel changes

* [FIX] A few Sentinel bugs fixed and improvements, see Redis 2.8.12
        changelog for more info.  (Salvatore Sanfilippo & Matt Stancliff)
* [NEW] New Sentinel-Client handshake protocol, ROLE command, CLIENT KILL,
        all backported to 3.0 branch. (Salvatore Sanfilippo)

--[ Redis 3.0.0 Beta 6 (version 2.9.55) ] Release date: 9 jun 2014

This is the 6th beta of Redis 3.0.0.

>> General changes

* [FIX] Fix software watchdog signal handler crash due to re-entering.
* [FIX] Better Lua number -> string conversion for Lua scripts.
* [FIX] Serious replication bug when min-slaves-* feature is used in slaves
        configuration fixed.
* [FIX] Blocking pop on lists now works when the list is created by commands
        other than *PUSH* (for example SORT STORE).

>> Cluster changes

* [FIX] CRITICAL: For a bug in the implementation of CLUSTER SET-CONFIG-EPOCH
        introduced with beta-3 (especially beta-4 where the command
        is actually used by redis-trib), a configEpoch for a node could
        jump backward, breaking the eventual consistency property of the
        slots -> nodes mapping in the cluster.

>> Sentinel changes

* No changes for Sentinel in this release.

--[ Redis 3.0.0 Beta 5 (version 2.9.54) ] Release date: 26 may 2014

This is the 5th beta of Redis 3.0.0. It does not include any real
worthwhile change (just three days passed since the previous beta), but
fixes two stupid bugs preventing cluster tests to pass.

--[ Redis 3.0.0 Beta 4 (version 2.9.53) ] Release date: 23 may 2014

This is the fourth beta of Redis 3.0.0.

>> General changes

* [NEW] Scripting engine performances improvements.
* [NEW] Log format slightly changed to report current node role.

* [FIX] Correct the HyperLogLog stale cache flag to prevent unnecessary
        computation.

>> Cluster changes

* [NEW] redis-trib: ability to import data from standalone Redis instances.
* [NEW] redis-trib: "fix" subcommand much better at fixing errors.
* [NEW] CLUSTER FAILOVER FORCE implemented.
* [NEW] CLUSTER RESET implemented, it is now possible to completely reset
        nodes and create a new cluster without restarting anything.
* [NEW] Slave validity factor (max estimated data age to still failover)
        is now under the control of the user, and can be configured via
        redis.conf or CONFIG SET. Option name cluster-slave-validity-factor.
* [NEW] Cluster test: failure detection and failover initial tests.
* [NEW] CLUSTER MEET: better error messages when address is invalid.
* [NEW] Bulk-accept new Cluster nodes in the Cluster bus instead of
        performing just a single accept() per event fired.

* [FIX] Bypass data_age check for manual failovers.
* [FIX] Fixed data_age computation / check integer overflow.
* [FIX] Various fixes to Tcl client.tcl Redis Cluster client used in tests.
* [FIX] Better handling of stolen slots.
* [FIX] Don't accept cluster bus connections during startup.

>> Sentinel changes

* [NEW] Generate +config-update-from event when a new config is received.
* [NEW] Log when a failover will be re-attempted again.

* [FIX] Sentinel: Add "dir /tmp" directive in example sentinel.conf.

--[ Redis 3.0.0 Beta 3 (version 2.9.52) ] Release date: 5 may 2014

This is the third beta of Redis 3.0.0.

>> General changes

* [NEW] New data structure: the HyperLogLog (see 2.8 release notes).
* [NEW] Lexicograhical range queries in sorted sets (see 2.8 release notes).
* [NEW] LRU algorithm precision greatly improved.

* [FIX] Redis is now much more responsive to reply with LOADING / BUSY errors.

>> Cluster changes

* [NEW] Cluster testing framework and initial tests.
* [NEW] Cluster epochs collision resolution (make Redis Cluster more resilient
        to admin and programming errors).
* [NEW] Persist / fsync some global state to ensure correct crash-recovery
        semantics.
* [NEW] New command introduced: CLUSTER SET-CONFIG-EPOCH, still not used
        by redis-trib. Will be used to speedup the assignment of unique
        epochs to different nodes at cluster creation time. For now this is
        handled as a side effect of the cluster epochs collision resolution.

* [FIX] Different fixes to redis-trib cluster creation.
* [FIX] Fix an error in the CLUSTER NODES output for nodes slots.

>> Sentinel changes

* [NEW] Sentinels are now able to send update messages in a peer-to-peer
        fashion even if no Redis instances are available. Now the Sentinel
        liveness property that the most updated configuration in a given
        partition is propagated to all the Sentinels is extended to partitions
        without reachable instances.
* [NEW] Sentinel safety properties are now ensured in a crash-recovery system
        model since some state is persisted on disk before replying to other
        nodes, and reloaded at startup.
* [NEW] Sentinel now uses CLIENT SETNAME so that it is easy to identify
        Sentinels using CLIENT LIST among other clients.
* [NEW] Sentinel failure detection and reconnection code improved.

--[ Redis 3.0.0 Beta 2 (version 2.9.51) ] Release date: 11 mar 2014

This is the second beta of Redis 3.0.0.

>> General changes

* [FIX] Sometimes the absolute config file path was obtained in a wrong way.
        This happened when there was a "dir" directive inside the config file
        and at the same time the configuration file was given as a relative
        path to redis-server or redis-sentinel executables.
* [FIX] redis-cli: Automatically enter --slave mode when SYNC or PSYNC are
        called during an interactive session.
* [FIX] BITCOUNT: fixed unaligned access causing issues in sparc and other
        archs not capable of dealing with unaligned accesses. This also makes
        the code faster in archs where unaligned accesses are allowed.
* [FIX] Force INFO used_memory_peak to match peak memory. This generated some
        confusion among users even if it was not an actual bug.
* [FIX] Fixed an critical EVALSHA script cache bug: scripts executed may not
        propagate to AOF / Slaves correctly under certain conditions.
        See issue #1549 at Github for more information.
* [FIX] Fixed multiple bugs resulting into closing the link with master or slave
        during replication without good reasons. This will result in useless
        resynchronizations, or infinite loops where the replication link can't
        be established.
* [FIX] Don't count the time needed to populate the buffers of clients waiting
        in MONITOR mode when populating the Slow Log entries.

* [NEW] Redis-cli updated to use SCAN instead of random sampling via
        RANDOMKEY in order to implement --bigkeys feature. Moreover the
        implementation now supports pipelining and reports more information
        at the end of the scan. Much faster, much better. A special thank
        you to Michael Grunder for this improvement.
* [NEW] redis-cli now supports a new --intrinsic-latency mode that is able
        to meter the latency of a system due to kernel / hypervisor.
        How to use it is explained at http://redis.io/topics/latency.
* [NEW] New command BITPOS: find first bit set or clear in a bitmap.
* [NEW] CONFIG REWRITE calls are now logged.
* [NEW] AOF write errors (like no space on device) no longer abort Redis if the
        fsync policy is none or every second. The database enters a read-only
        mode where every write is refused with an error. Normal operations are
        restored as soon as Redis is able to append again data to the AOF file.
* [NEW] Sentinel now accepts SHUTDOWN command.


>> Cluster changes

* [FIX] Bind the first interface listed in the "bind" configuration directive
        if any, in order to perform outgoing connections. This fixes Cluster
        usage when an address is bound but there are multiple interfaces that
        may be used to connect with other nodes.
* [FIX] When an "Importing" slot is closed via CLUSTER SETSLOT NODE ...
        increment the configEpoch in the special case it is zero.
* [FIX] Current transaction is invalidated on redirection errors.
* [FIX] Abort if port does not allow for a valid cluster bus port that is
        always at fixed +10000 offset.
* [FIX] Keys extraction algorithm fixed for ZUNIONSTORE/ZINTERSTORE and SORT.
* [FIX] Better failover timeout and retry times: failover should now work
        reliabily when node-timeout is very small (a few milliseconds).
* [FIX] Don't allow SORT GET/BY options in Cluster mode.
* [FIX] Clear importing/migrating state when turning from master to slave role.
* [FIX] Set slot error if we receive an update for a busy slot.
* [FIX] Update node configEpoch on UPDATE messages.

* [NEW] Support multi-key operations as long as keys resolve to the same
        hash slot, and the slot is not migrating, or it is migrating but all
        the mentioned keys are available.
* [NEW] New DEBUG command CMDKEYS available to debug / test keys identification
        in Redis commands.
* [NEW] redis-trib: create subcommand is now able to assign spare slaves.
* [NEW] redis-trib: new subcommand 'call'. Exec command in all nodes.


>> Sentinel changes

* [FIX] Sentinel "IDONTKNOW" error removed as it does not made sense with the
        new Sentinel design. This error was actually a fix for a design error
        in the first implementation of Sentinel.
* [FIX] Sentinel: added a missing exit() call to abort after config file
        checks at startup. This error was introduced with an improvement in
        a previous 2.8 release.
* [FIX] Sentinel: better nodes fail over start time desynchronization to avoid
        split-brain during the voting process needed to get authorization to
        fail over. This means the system is less likely to need to retry
        and will fail over faster. No changes in behavior / correctness.

* [NEW] Sentinel unit tests and framework. More tests needed and units must
        be improved in order to have less false positives, but it is a start
        and features a debugging console that is useful to fix tests or to
        inspect bugs causing tests failures.
* [NEW] New Sentinel events: +/-monitor and +set used to monitor when an
        instance to monitor is added or removed, or when a configuration
        is modified via SENTINEL SET.

--[ Redis 3.0.0 Beta 1 (version 2.9.50) ] Release date: 11 Feb 2014

This is the first beta of Redis 3.0.0.

Migrating from 2.8 to 3.0
=========================

Redis 2.8 is mostly a strict subset of 3.0, you should not have any problem
upgrading your application from 2.8 to 3.0. However this is a list of small
non-backward compatible changes introduced in the 3.0 release:

* The log format was modified. The prefix of each line included the pid
  in the following format [1234]. Now instead it is 1234:? Where
  '?' is actually the role of the instance. M for master, S for slave, C
  if this process is a saving child (for RDB/AOF), and X for Sentinel.

* The default maxmemory policy in Redis 3.0 is no longer "volatile-lru" as
  it used to be in 2.8, but "noeviction". The policies behavior is the same
  (but LRU eviction is much more precise in 3.0), so only the default value
  changed. Just make sure to specify in your redis.conf what you mean.

--------------------------------------------------------------------------------

Credits: Where not specified the implementation and design is done by
Salvatore Sanfilippo. Thanks to Pivotal for making all this possible.
Also many thanks to all the other contributors and the amazing community
we have.

See commit messages for more credits.

Cheers,
Salvatore