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 3.0.7 ] Release date: 28 jan 2016

Upgrade urgency MODERATE: this release fixes important Redis Cluster bugs.

* [FIX] avg_ttl reporting in INFO improved. (Salvatore Sanfilippo)

* [FIX] Redis Cluster address update (via gossip section) processing improved

to avoid initiating inwanted handshakes.

* [FIX] Many fixes to MIGRATE multiple keys implementation. The command

could handle errors in a faulty way leading to crashes or other

unexpected behaviors. MIGRATE command refactoring.

(The analysis of the faulty conditions was conducted by

Kevin McGehee. The fix was developed by Salvatore Sanfilippo)

* [FIX] A Redis Cluster node crash was fixed because of wrong handling of

node->slaveof pointers.

(Reported by JackyWoo, fixed by Salvatore Sanfilippo)

* [FIX] Fix redis-trib rebalance when nodes need to be left empty because

the specified weight is zero.

(Reported by Shahar Mor, fixed by Salvatore Sanfilippo)

* [FIX] MIGRATE: Never send -ASK redirections for MIGRATE when there are

open slots. Redis-trib and other cluster management utility must

always be free to move keys between nodes about open slots, in order

to reshard, fix the cluster configuration, and so forth.

(Salvatore Sanfilippo)

* [FIX] Redis-trib is now able to fix more errors. A new CLUSTER subcommand

called BUMPEPOCH was introduced in order to support new modes

for the "fix" subcommand. (Salvatore Sanfilippo)

* [NEW] Cluster/Sentinel tests now use OSX leak to perform leak detection

at the end of every unit. (Salvatore Sanfilippo)

* [NEW] Detect and show server crashes during Cluster/Sentinel tests.

(Salvatore Sanfilippo)

* [NEW] More reliable Cluster/Sentinel test becuase of timing errors and

-LOADING errors. (Salvatore Sanfilippo)

--[ Redis 3.0.6 ] Release date: 18 Dec 2015

Upgrade urgency: MODERATE. We fixed a crash that happens very rarely, so

updating does not hurt, but most users are unlikely to

experience this condition because it requires some odd

timing. However if you are a Redis Cluster user, upgrading

is strongly adviced since this release includes very

important improvements to Redis Cluster.

* [FIX] lua_struct.c/getnum security issue fixed. (Luca Bruno discovered it,

patched by Sun He and Chris Lamb)

* [FIX] Redis Cluster replica migration fixed. See issue #2924 for details.

(Salvatore Sanfilippo)

* [FIX] Fix a race condition in processCommand() because of interactions

with freeMemoryIfNeeded(). Details in issue #2948 and especially

in the commit message d999f5a. (Race found analytically by

Oran Agra, patch by Salvatore Sanfilippo)

* [NEW] Backported from the upcoming Redis 3.2:

MIGRATE now supports an extended multiple-keys pipelined mode, which

is an order of magnitude faster. Redis Cluster now uses this mode

in order to perform reshardings and rebalancings. (Salvatore Sanfilippo)

* [NEW] Backported from the upcoming Redis 3.2:

Redis Cluster has now support for rebalancing via the redis-trib

rebalance command. Demo here:

https://asciinema.org/a/0tw2e5740kouda0yhkqrm5790

Official documentation will be available ASAP. (Salvatore Sanfilippo)

* [NEW] Redis Cluster redis-trib.rb new "info" subcommand.

* [NEW] Redis Cluster tests improved. (Salvatore Sanfilippo)

* [NEW] Log offending memory access address on SIGSEGV/SIGBUS (Salvatore

Sanfilippo)

--[ 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