Repository navigation
v1.2.1
KiviDB v1.2.1
Upgrade to 1.2.2. It fixes GHSA-8rm8-p5c5-g3mr, which affects 1.2.0 and 1.2.1: with a search index, a multi-key write over an expired key could stop the server answering.
What's in 1.2.1
1.2.1 is a patch release of 1.2.0. It fixes the 1.2.0 known issues of reshards that could lose writes and replicas that could stop while they saved, and many smaller ways in which a replica, the append-only file or a reshard could end up with other data than the master:
- Clusters.
- A reshard under load no longer loses acknowledged writes, with
MIGRATE(redis-cli --cluster reshard) or withCLUSTER MIGRATION IMPORT(the Kubernetes operator): a write checks inside the write gate whether its keys have left the node, and is sent on with-ASKor-MOVEDif they have. The same holds forEXECand scripts, and for every other way a slot leaves a node (#331, #337, #339, #407). - A failover of the source during an atomic slot migration no longer loses the writes the new master acknowledged: a failover and a handover now exclude each other (#516).
- A drained node no longer deletes the keys it moved from its new copy of them (#348), keys written to a slot that came back are no longer deleted with its old keys (#353), and writes no longer stay paused after a handoff (#349).
MIGRATEno longer stopsredis-cli --cluster reshardwithTRYAGAINfor an unchanged hash whose field is about to expire (#332).- Cluster admin commands (
CLUSTER FLUSHSLOTS,ADDSLOTS,ADDSLOTSRANGE,BUMPEPOCH,SET-CONFIG-EPOCH,COUNT-FAILURE-REPORTS,GETKEYSINSLOT) answer as in Redis (#552, #553, #563), and a restarted node gives its own address at once (#307).
- A reshard under load no longer loses acknowledged writes, with
- Replicas and the AOF stay in step with the master.
- A replica no longer stops for good while it saves or is promoted (#373).
CLIENT PAUSE ... WRITEreplies only once every write already being applied is in the offset, so a switchover (the operator's, and any that works the same way) no longer promotes a replica without the last acknowledged write (#629).- A write to a key past its TTL first deletes it and sends its
UNLINKon, so a replica no longer keeps a key the master no longer has (#628). A write that changed nothing is no longer sent on, as in Redis. - A hash field past its TTL is removed by the master only, and sent on as an
HDEL(#415). Its deadline is kept to the millisecond everywhere, and a replica holds the master's (#594, #489). A relative TTL reaches replicas and the AOF as the master's own deadline (#602). - A write that fails changes nothing (
JSON.MSET,XGROUP CREATE ... MKSTREAM,TS.ADD) (#551),SPOPwith a count from a RESP3 client is replicated, and a write in database 0 no longer reaches replicas and the AOF in another database (#423). - A replica that falls behind no longer makes its master evict its keys and refuse writes (#542).
- A master that becomes a replica ends its blocked clients, as Redis does (#591). On
--io-model tokio,CLIENT KILLends a replication link (#595).WAITandWAITAOFno longer starve socket I/O on io_uring (#498). - Sentinel no longer promotes a replica set to
replica-priority 0, and new--replica-priorityand--replica-announcedflags set them (#443).
- Durability.
appendfsync alwayssyncs each write to disk before it is acknowledged, as in Redis (#513). It was accepted and did nothing.- A restart with the AOF on after a save with it off no longer applies writes twice (#354), and
CONFIG SET appendonly yesstarts a new log instead of appending to an old one (#350, #368, #359). A log that is not replayed is never deleted: it is kept aside indirassuperseded.<time>.<n>.appendonly.aof, and counted inINFO persistence(aof_superseded_files,aof_superseded_bytes). - With
appendonly no, noappendonly.aofis made or needed (#47). - A snapshot or RDB file that is not whole (cut short, a wrong checksum, a Dragonfly
.dfsfile, databases the server does not have) is refused instead of loaded in part (#444, #474, #477, #482).
- Faster where 1.2.0 was slow.
INFOno longer stalls writes on a server with many keys that have a TTL: about 0.1 ms with 5 million of them, against 55 ms (#622).- An idle server with many keys that have a long TTL no longer spends CPU on them: 0.5-0.75% of a core with 5 million keys, against about 40% (#618).
HSETno longer takes a lock every worker shares when no search index is defined (#544).ZREM,ZCOUNTandZREMRANGEBYSCOREno longer scan the whole sorted set (#386, #408).
- Moving data in. KiviDB reads Valkey 9.0's and 9.1's RDB format: a
dump.rdb,--preload-file, a full sync from a Valkey 9 primary andRESTOREof itsDUMPpayloads (#449). Dragonfly's hashes with field TTLs and JSON documents, KeyDB's field TTLs andDUMPpayloads, and Redis Stack 6.2's files load; what KiviDB does not read is refused by name (#452). An RDB file's databases past 15 load into their own databases (#477). - Keys are bytes, as in Redis. Every command takes a key that is not UTF-8, as do stream group and consumer names, vector-set elements,
KEYSandSCANpatterns and passwords (#437, #446). - Compatibility.
HELLOreports the Redis release KiviDB follows (server: redis,version: 8.10.0), so redis-py and Jedis turn on their client-side caches (#451).- A client that goes while it is blocked no longer takes the next element, which was lost with Bull and n8n's queue mode (#384).
BRPOPLPUSHandBLMOVEtime out with*-1(#385). - A Lua script's
falseis a RESP3 null to a RESP3 client, so Celery chords complete with redis-py 8 (#431). @connectionholds what it holds in Redis, so redis-py's cluster client connects with+@connection(#392).MULTI,WATCH,HELLO, the ACL log,TTLrounding, expire-time bounds, blocking commands in scripts,SWAPDB,DEBUG, cluster, JSON, search and time series commands answer as Redis, RedisJSON, RediSearch and RedisTimeSeries do. JSON numbers keep every digit (#445).
- Installing.
- Signed apt and dnf repositories at
https://packages.kividb.io, foramd64andarm64:apt install kividbanddnf install kividb. See docs/INSTALL.md. - The
.deband.rpmsay they need glibc 2.34. - The release is published on crates.io:
cargo install --locked kividb, orcargo binstall kividb. - Every release has a source tag,
X.Y.Z-src, and source archives are small again.
- Signed apt and dnf repositories at
CHANGELOG.md has the full list.
Security
1.2.1 fixes two issues; upgrade from any 1.1.x or 1.2.0.
- Commands that write only with some arguments are checked as writes (GHSA-cg4w-p5w5-57w5, high).
GEORADIUS/GEORADIUSBYMEMBER ... STORE|STOREDISTandMIGRATEwere let through by read-only key permissions such as%R~*.+@readgranted them, and-@writedidn't remove them or theFUNCTION LOAD|DELETE|FLUSH|RESTOREsubcommands.min-replicas-to-write, the out-of-memory refusal andaof-ack-after-writedidn't apply to them.- They are now refused with
NOPERM,NOREPLICASorOOM, as in Redis. - Affects 1.0.0 to 1.2.0.
- Deeply nested JSON documents could crash the server (GHSA-qh83-449g-hwm5, medium).
- Repeated
JSON.SETat ever deeper paths could build a document thousands of levels deep. Reading or copying it then overflowed the stack and stopped the process. - A write that would nest a document 128 levels or deeper is now refused with
ERR recursion limit exceeded, as RedisJSON does. - Snapshots and
RESTOREpayloads holding such a document, or an array index past the largest allowed, are refused. - Affects 1.1.0 to 1.2.0.
- Repeated
Before you upgrade
- Upgrade replicas before their master, as always. 1.2.1 can write what 1.2.0 cannot read:
- A 1.2.0 server refuses to start on a snapshot that holds a key, a consumer group or consumer name, or a vector-set element name that is not UTF-8, or the empty key, and says so. A 1.2.0 replica cannot full-sync from a 1.2.1 master that holds one, and drops a replicated write of such a key. A 1.2.0 node refuses such a key in
RESTORE, soMIGRATEto it fails and the source keeps the key. - A 1.2.1 replica of a 1.2.0 master is fine.
- A 1.2.0 server refuses to start on a snapshot that holds a key, a consumer group or consumer name, or a vector-set element name that is not UTF-8, or the empty key, and says so. A 1.2.0 replica cannot full-sync from a 1.2.1 master that holds one, and drops a replicated write of such a key. A 1.2.0 node refuses such a key in
- No going back to 1.2.0 once 1.2.1 has saved such a key. Take a backup before you upgrade if you may need to go back.
- Hash fields expire on replicas when the master says so. A replica, and the AOF as it is replayed, keep a field past its TTL until the master's
HDEL; a replica hides it from its readers meanwhile.- While any node of a master and its replicas is 1.2.0, fields are still removed by each node's own clock. The gap closes once the master, upgraded after its replicas, is 1.2.1.
- An AOF written by 1.2.0 holds no
HDELfor the fields that expired. RunBGREWRITEAOFonce after upgrading. - A
SORTwhoseBYorGETpattern reads a hash field now runs alone, as a script does.
WAITwith no replicas waits out its timeout, as Redis does:WAIT n timeoutwithnabove 0 answered at once, and now answers 0 aftertimeout.WAITis refused on a replica, and a negative timeout is refused.WAIT 0answers the number of replicas that have what was written, not the number connected.HELLOsaysserver: redisandversion: 8.10.0. A tool that told KiviDB apart byHELLO'sservershould look forkividb_versioninstead, as inINFO.GEORADIUSandGEORADIUSBYMEMBERare refused on a read-only replica, with or withoutSTORE, as in Redis. An application that reads from replicas with them must switch toGEORADIUS_ROandGEORADIUSBYMEMBER_RO, orGEOSEARCH.- ACL categories are Redis's, and apply to existing users as soon as the server is upgraded:
- A user with
+@writebut not+@connection(or+@all) can no longerSELECT. Add+select. - A user with
+@read,+@keyspaceor+@genericbut not+@connectioncan no longer runWAITorWAITAOF. Add+wait +waitaof. - A user of
+@all -@write, or of thereadonlyshorthand, may nowSELECT.
- A user with
ACL LOGanswers as Redis does: fields in Redis's order,entry-id(notid) counting from 0, and the ten newest entries without a count. A client that reads entries by position or byidneeds to change.acllog-max-len(default 128) is now applied.CONFIG REWRITEneeds a config file. A server started without one now repliesERR The server is running without a config file. To use it in the Docker image, start the server with--configfilepointing at a file that exists. On a host where an earlier version created/etc/kividb/kividb.confthis way, look for areplicaofline nobody wrote, and delete or fix it.- Delete sorted sets with a NaN score before upgrading. No command makes one, but a 1.2.0 server holding one (from a crafted
RESTOREor snapshot) won't start on 1.2.1 if the key is in its snapshot or AOF preamble.ZRANGE key 0 -1 WITHSCORESshowsnan. - A search index is seen only from its own database.
FT._LISTlists the selected database's indexes, and the commands that name an index of another database replyIndex not found. AnFT.ALTER,FT.ALIASADD,FT.ALIASDEL,FT.ALIASUPDATEorFT.SYNUPDATEthat a 1.2.0 AOF or master holds for another database's index is refused on replay. Once the master is upgraded, check such an index withFT.INFOfrom its own database and repeat what is missing. - Time series.
DUMP,RESTOREandMIGRATEleave out a series' compaction rules and source, as in RedisTimeSeries. An atomic slot migration from a 1.2.0 node, as in a rolling upgrade, loses the moved series' rules: make them again withTS.CREATERULEafterwards.- The options take RedisTimeSeries's spellings (
CHUNK_SIZE), and a word that is not an option no longer drops the options after it. A 1.2.0 AOF replays whole, under these rules, so after the first restart a series can differ from what 1.2.0 held in memory. - A sample's time above
i64::MAXis refused.
JSON.NUMINCRBYandJSON.NUMMULTBYkeep a float result a float (1.5 + 1.5is3.0), and refuse an integer overflow withERR numeric overflow. A client that reads such a result as an integer may need to change.- Expire times are checked as Redis checks them. A deadline past
i64::MAXmilliseconds isERR invalid expire time, a number too large isERR value is not an integer or out of range,EXAT 0andPXAT 0are refused, andSETwithKEEPTTLand a time, or with two ofEX,PX,EXATandPXAT, is a syntax error. A hash field's deadline must be from 0 to 2^46 - 1 ms after the epoch.TTLandEXPIRETIMEround to the nearest second. - In
MULTI,UNWATCH,AUTH,HELLO,MONITOR,WAITandWAITAOFare queued, as in Redis 8.MONITORis refused there, andWAITandWAITAOFanswer at once. appendfsync alwaysnow costs writes a second, as in Redis.everysecstays the default.CLIENT PAUSEcan answer an error (writes in progress did not finish before the pause could take hold) when a write being applied takes longer than 5 seconds, and then pauses nothing.- During an atomic slot migration's handoff, and for 3 s after it,
CLUSTER FAILOVERon a replica of the source repliesOKbut does not happen. The source and the replica log why. Ask again a few seconds later. - Replication links in
CLIENT LISTshowflags=Son the master andflags=Mon the replica, andconnected_clientsno longer counts them. - The packages need glibc 2.34 or newer: Debian 12, Ubuntu 22.04, RHEL 9, Amazon Linux 2023 or later. On Debian 11, Ubuntu 20.04, RHEL 8 or Amazon Linux 2 the package manager refuses them.
Known issues
- AOF: a power loss during a checkpoint can leave a gap in the log (#277), as in 1.2.0. Only a power loss or a kernel crash is affected.
- Memory freed by dropping a large vector index stays with the process until
MEMORY PURGE, as in 1.2.0. redis-cli --cluster reshardunder heavy writes slows the source node's writes to seconds (#310), as in 1.2.0. No write is lost.CLUSTER MIGRATION IMPORT, which the Kubernetes operator uses, isn't affected.
Benchmarks
benchmarks/1.2.1 measures 1.2.1 against the 1.2.0 release binary, with the 1.2.0 suites' method and harnesses: throughput and latency by data type, memory per key, instructions per command, vector search, and memory during a snapshot. benchmarks/1.2.0 has the comparison with other servers.
Thanks
Thanks to the contributors whose work is in this release:
@Parikalp-Bhardwaj (hash fields expired by the master, MEMORY USAGE of sets, hashes and sorted sets, ZREM, ZCOUNT and ZREMRANGEBYSCORE without a scan, active defrag's relocation made safe against a panic, and a test of PFMERGE and PFADD),
@Ayege (JSON.NUMINCRBY and JSON.NUMMULTBY floats and overflow),
@matheussilvano (SCRIPT HELP, latency-tracking-info-percentiles, INFO commandstats, and the LangChain, LangGraph, LlamaIndex and Mem0 harnesses),
@paddor (WAIT and WAITAOF on io_uring),
@Sourabh-ctrl (SCORE and SCORE_FIELD as in Redis),
@saikishore222 (DEBUG),
@bfs2021 (the eviction tests), @ZYZ666-RGB (the LlamaIndex harness) and @Anandb71 (the chat compatibility checks).
Binary variants
Each OS/architecture ships four binaries — pick the one that matches your needs:
| Tarball suffix | Features included | When to use |
|---|---|---|
| (none) | — | Minimal / constrained environments |
-lua |
Lua scripting | EVAL / EVALSHA |
-tls |
TLS termination | Encrypted client connections |
-full |
Lua + TLS | Recommended general-purpose build |
Full changelog: CHANGELOG.md
License: GNU Affero General Public License v3 (AGPL-3.0-only)
Code of conduct: CODE_OF_CONDUCT.md
Changes since v1.2.0:
What's Changed
- Release notes 1.2.0: the security fixes by @murtazayusuf in #316
- Update README.md by @murtazayusuf in #317
- Source archives leave out the raw benchmark results (#319) by @murtazayusuf in #321
- 1.2.0 docs: changes against 1.1.3 only; README and the benchmark report by @murtazayusuf in #320
- README: the 1.2.0 benchmark video and charts by @murtazayusuf in #324
- Release: push an X.Y.Z-src tag with each stable release (#319) by @murtazayusuf in #322
- README: the 1.2.0 benchmark video and charts by @murtazayusuf in #326
- LangChain compatibility: a harness, its results against KiviDB and Redis 8, and a docs page by @matheussilvano in #325
- LangGraph compatibility: a harness, its results against KiviDB and Redis 8, and a docs page by @matheussilvano in #329
- LangChain harness: remove the indexes it creates by @matheussilvano in #330
- Add SCRIPT HELP by @matheussilvano in #318
- CONFIG SET latency-tracking-info-percentiles refuses what Redis refuses by @matheussilvano in #323
- Release: don't open the Announcements discussion from the workflow by @murtazayusuf in #328
- Release notes 1.2.0: known issue #331 (reshard with MIGRATE under load can lose writes) by @murtazayusuf in #333
- Release notes 1.2.0: #331 affects CLUSTER MIGRATION IMPORT too by @murtazayusuf in #336
- Fix lost writes when a write or EXEC races a slot move (#331), and a spurious TRYAGAIN for hashes (#332) by @murtazayusuf in #335
- LlamaIndex harness: Redis vector store, docstore, index store and kv store by @matheussilvano in #334
- LlamaIndex harness: drop only this run's indexes, and correct the report by @matheussilvano in #341
- LangGraph harness: drop only this run's indexes by @matheussilvano in #343
- Mem0 harness: Redis vector store and the Memory API by @matheussilvano in #342
- Cluster: a restarted node keeps the address nodes.conf has for it (#307) by @murtazayusuf in #344
- Route a write again in the gate after every way a slot leaves a node (#337) by @murtazayusuf in #345
- With appendonly no, no appendonly.aof is made or needed by @murtazayusuf in #346
- test: cover PFMERGE and PFADD lost-update race by @Parikalp-Bhardwaj in #312
- Cluster: a drained source that followed the destination keeps its keys (#348); a FAIL hold no longer leaves a handoff's pause on (#349) by @murtazayusuf in #351
- fix(compaction): make entry relocation unwind-safe by @Parikalp-Bhardwaj in #356
- Replay the AOF only over the dump it goes on from; start a new log on a runtime enable by @murtazayusuf in #352
- Cluster: a deletion of a lost slot's keys keeps what was written to it since it came back (#353) by @murtazayusuf in #355
- AOF: follow-ups from #352 (dir sync, logs added to after a save, partial kept copies, INFO, docs) by @murtazayusuf in #360
- AOF: review follow-ups to #360 (old log added to by a turn-over, log beside an older old log, kept-log warnings, kubectl recipe) by @murtazayusuf in #362
- Search: a query reads the watched hashes before the postings (#347); hostname test waits for nodes.conf (#338) by @murtazayusuf in #361
- AOF: judge the log on its own beside an old log older than the dump (#363) by @murtazayusuf in #365
- Active defrag: a pass that panics is caught, its slabs put back, and the thread goes on (#358) by @murtazayusuf in #366
- Update FUNDING.yml with Patreon username by @murtazayusuf in #369
- Field expiry: sweep index and watch updates under the write order; TEXT queries try watched hashes (#364) by @murtazayusuf in #367
- AOF: keep a stopped turn-over's old log after a save with the AOF off (#368) by @murtazayusuf in #372
- AOF: follow-ups from #372 (no needless startup fatal, message text, keep-aside before the log lock) by @murtazayusuf in #377
- Replica: applied writes take their keys' order while a hash is watched; TEXT fallback reads the watched list only for words (#371) by @murtazayusuf in #376
- Active defrag panics: review follow-ups to #366 (#370) by @murtazayusuf in #374
- AOF: #377 review nits (warning says what to do, no second kept copy) by @murtazayusuf in #379
- Compatibility: LightRAG, cognee and OmniRoute on KiviDB 1.2.0 by @murtazayusuf in #389
- Dify and RAGFlow on KiviDB: end-to-end harnesses, results and guides by @murtazayusuf in #394
- Compat: claude-mem, OpenViking and Mem0's TypeScript SDK on KiviDB 1.2.0 by @murtazayusuf in #395
- Compatibility list: Dify and RAGFlow by @murtazayusuf in #397
- README: Trendshift badge by @murtazayusuf in #398
- n8n and Firecrawl compatibility: harnesses, results and guides by @murtazayusuf in #396
- README: banner in kividb.io's style by @murtazayusuf in #400
- Replica: enter the command gate before the apply lock (#373); flushes and SWAPDB from the stream take it alone (#378) by @murtazayusuf in #382
- mem0 report: the TypeScript SDK is covered by mem0-ts-compat now by @murtazayusuf in #399
- BRPOPLPUSH and BLMOVE time out with a null array, as in Redis (#385) by @murtazayusuf in #404
- io_uring: let a blocked client go as soon as it closes, and never give it an element (#384) by @murtazayusuf in #403
- Close the last gaps around a woken client that goes (#384 follow-ups) by @murtazayusuf in #405
- Cluster: EXEC during a handoff is refused only if it may write (#339) by @ShryukGrandhi in #406
- perf(zset): eliminate linear scan from ZREM by @Parikalp-Bhardwaj in #393
- RAGFlow 1.0 on KiviDB: harness, RAGFlow's Go suite, results and guide by @murtazayusuf in #410
- ACL: @connection holds what it holds in Redis (#392) by @murtazayusuf in #409
- Fix SET-ACTIVE-EXPIRE, subcommand errors, and OBJECT key display (#236) by @saikishore222 in #380
- fix(search): match Redis document score compatibility by @Sourabh-ctrl in #296
- perf(zset): use bounded traversal for score-range commands by @Parikalp-Bhardwaj in #412
- Cluster: GEORADIUS … STORE in a transaction is a write for the pause and the handoff (#407) by @murtazayusuf in #416
- Score-range bounds are read as Redis reads them by @murtazayusuf in #418
- Refuse a NaN sorted-set score when restoring or loading by @murtazayusuf in #419
- SELECT is not refused or held as a write by @murtazayusuf in #417
- Update GitHub Sponsors username in FUNDING.yml by @murtazayusuf in #427
- The first SELECT of another database takes its cut before the write gate by @murtazayusuf in #425
- Compatibility: LiteLLM, Celery, Semantic Kernel, Spring AI, GPTCache, Upstash Ratelimit and Haystack on KiviDB 1.2.0 by @murtazayusuf in #435
- COMPATIBILITY: keys, hash fields and set members must be UTF-8 (#437) by @murtazayusuf in #438
- README: Hacker News badge by @murtazayusuf in #440
- Docs: HyperLogLogs from Redis load and count right; Valkey 9 and copy-back gaps; UTF-8 key names by @murtazayusuf in #454
- Lua: a script's false is the null of the client's protocol (#431) by @murtazayusuf in #436
- Keys as bytes in the store: KeyBytes, and the store API over key bytes (#437, 1 of 4) by @murtazayusuf in #460
- JSON: read numbers exactly, so every double round-trips (#445) by @murtazayusuf in #463
- Command tables: regenerate from Valkey's BSD command data by @murtazayusuf in #466
- Publish to crates.io: cargo install kividb and cargo binstall kividb by @murtazayusuf in #457
- Signed apt and dnf repositories at packages.kividb.io by @murtazayusuf in #459
- CONFIG REWRITE: refuse when the server was not started from a config file (#442) by @murtazayusuf in #456
- Valkey 9's RDB format loads: files, full sync, RESTORE (#449) by @murtazayusuf in #461
- Snapshots: the empty key no longer ends its database (#441) by @murtazayusuf in #455
- LlamaIndex compatibility: re-run the harness on current master, still 54 of 54 by @ZYZ666-RGB in #480
- FT._LIST lists the selected database's indexes, and an index of another database is not found by @murtazayusuf in #465
- RDB: keep databases past 15 apart, or refuse the file (#477) by @murtazayusuf in #479
- A hash field's deadline is converted to the nanosecond when a hash is saved and loaded by @murtazayusuf in #467
- replica-priority and replica-announced reach INFO replication, so Sentinel never promotes a replica of priority 0 by @murtazayusuf in #469
- Snapshots: refuse a KiviDB image with keys past the server's databases (#482) by @murtazayusuf in #487
- A negative Unix time for a hash field is refused, as by Redis, and the fields are kept by @murtazayusuf in #488
- RDB files from Redis Stack 6.2 load: read past RedisGraph's module data by @murtazayusuf in #485
- Update Hacker News badge in README by @murtazayusuf in #491
- fix(memory): accurate MEMORY USAGE for sets, hashes and sorted sets by @Parikalp-Bhardwaj in #492
- Dragonfly and KeyDB data loads whole or is refused by name (#452) by @murtazayusuf in #464
- Keys as bytes everywhere the keyspace is read back (#437, 2 of 4) by @murtazayusuf in #490
- Fix the master build: FieldClock for the Dragonfly and KeyDB field-TTL readers by @murtazayusuf in #495
- Fix the master build: Key keys in the RDB loader's KeyDB and Dragonfly code by @murtazayusuf in #496
- Snapshots: refuse a file that is not a whole snapshot (#444) by @murtazayusuf in #462
- Update Hacker Git badge style in README by @murtazayusuf in #501
- io_uring: prevent WAIT from starving replica I/O by @paddor in #498
- HELLO reports the Redis release KiviDB follows, so redis-py and Jedis turn on client-side caching by @murtazayusuf in #468
- Snapshot: refuse an image cut short inside a section after its databases (#474) by @murtazayusuf in #503
- Tests: the MEMORY USAGE tests probe ports without listening on them (#506) by @murtazayusuf in #507
- Keys, stream names and vector-set elements of any bytes (#437, step 3) by @murtazayusuf in #504
- Tests: read the eviction counters with the limit lifted by @bfs2021 in #511
- appendfsync always: sync each write before its reply (#513) by @murtazayusuf in #514
- CHANGELOG: the #447 upgrade note lists FT.ALIASDEL and FT.SYNUPDATE, not FT.DROPINDEX by @murtazayusuf in #522
- CLIENT PAUSE WRITE does not hold SCRIPT LOAD or SCRIPT FLUSH (#518) by @murtazayusuf in #525
- memory_usage_accuracy: hold the shrink sweep off while the other keys are deleted (#508) by @murtazayusuf in #526
- Cluster: a failover of the source during an atomic slot migration loses no acknowledged write (#516) by @murtazayusuf in #523
- AOF: a sync of an old log clears only the turn-over it found (#529) by @murtazayusuf in #530
- fix(field-expiry): make hash-field expiration master-driven by @Parikalp-Bhardwaj in #531
- CHANGELOG: patterned SORT and lua-parallel-scripts (#415) by @murtazayusuf in #532
- INFO commandstats counts WATCH, UNWATCH, HELLO, DISCARD, RESET and AUTH (#388) by @matheussilvano in #533
- JSON.NUMINCRBY keeps a float result a float (#470) by @Ayege in #537
- Tests: another test's load or pause no longer stops a field-expiry test's expiry (#535) by @murtazayusuf in #538
- JSON.NUMINCRBY refuses an integer overflow (#536) by @Ayege in #540
- Benchmarks: use gpt-6-luna for chat compatibility checks by @Anandb71 in #541
- Benchmarks: 1.2.1 against 1.2.0 by @murtazayusuf in #547
- CHANGELOG, COMPATIBILITY: hash-field expiry in both mixed-version directions; one upgrade order by @murtazayusuf in #548
- HSETEX builds its replicated form only when something reads it (#543) by @murtazayusuf in #549
- HSET: ask whether any search index exists without taking the indexes' lock (#544) by @murtazayusuf in #569
- Queue UNWATCH, AUTH, HELLO, MONITOR, WAIT and WAITAOF inside MULTI by @murtazayusuf in #573
- parity tests: commandstats counts are not disturbed by other tests by @murtazayusuf in #580
- Cluster admin commands answer as Redis does (#552, #553, #563) by @murtazayusuf in #577
- Arrays, JSON and search replies as Redis gives them; JSON events only on change (#554, #567, #575) by @murtazayusuf in #581
- A write that fails changes nothing: JSON.MSET, XGROUP CREATE MKSTREAM, TS.ADD (#551) by @murtazayusuf in #571
- TTL, PTTL, EXPIRETIME and HEXPIRETIME round as Redis does by @murtazayusuf in #574
- MULTI follow-ups: PSYNC/SYNC refused, ACL-denied EXEC discards, arity before ACL, HELLO protocol after AUTH by @murtazayusuf in #585
- Scripts, transactions and SWAPDB: blocking pops in scripts, SWAPDB watches, FUNCTION FLUSH order by @murtazayusuf in #578
- JSON.ARRINDEX, FT.SPELLCHECK, ARSCAN LIMIT and FT.CONFIG HELP as RedisJSON/RediSearch 8.10 (#583) by @murtazayusuf in #587
- TS options: CHUNK_SIZE, unknown words skipped, TS.INCRBY/DECRBY create with their options by @murtazayusuf in #579
- Redis differences in arguments, replies and error texts (#561, #562, #564, #566, #576) by @murtazayusuf in #582
- CI: time limit and retries on the client-dependency installs by @murtazayusuf in #597
- A lagging replica no longer makes its master evict its keys and refuse writes (#542) by @murtazayusuf in #572
- HELLO as Redis's helloCommand; no-multi and unknown commands before ACL; refused EXEC to MONITOR by @murtazayusuf in #596
- Refuse expire times past the maximum, as Redis does by @murtazayusuf in #589
- WAIT follow-ups: hub-level tests of the counting rules, one count per replica by @murtazayusuf in #603
- TS: refuse a sample time above i64::MAX; compaction links left out of DUMP/RESTORE and checked where used by @murtazayusuf in #599
- CONFIG SET appendonly yes off the runtime; fix three flaky tests by @murtazayusuf in #604
- README: run the release binary from the kividb/ directory the tarball unpacks to by @murtazayusuf in #608
- replication tests: wait loops take a LOADING reply as not yet by @murtazayusuf in #607
- End blocked clients on a role change; CLIENT KILL ends replication links (#591, #595) by @murtazayusuf in #600
- WATCH and client-side caching: only the keys a write changed by @murtazayusuf in #598
- Keep hash field deadlines as Unix milliseconds: exact round trips, replicas hold the master's deadline (#594, #489) by @murtazayusuf in #601
- README: commands and claims that did not hold when run by @murtazayusuf in #609
- FT.CREATE: add the first index once the writes that found none are done (#570) by @murtazayusuf in #586
- binary_keys tests: wait until the dataset is loaded by @murtazayusuf in #610
- Send the stored deadline for relative TTLs of SET, SETEX, MSETEX and RESTORE (#602) by @murtazayusuf in #606
- README: link the independent reviews by @murtazayusuf in #614
- README: lead with the latest benchmark numbers by @murtazayusuf in #615
- README: lead with the 1.2.0 bare-metal numbers by @murtazayusuf in #616
- benchmarks/1.2.1: re-run on the final commit 7407153 by @murtazayusuf in #617
- Active expiry looks only at keys that may be due (#618) by @murtazayusuf in #623
- Expiry: bring a far horizon back after a clock step; stricter model test (follow-up to #623) by @murtazayusuf in #625
- Version 1.2.1-rc1 by @murtazayusuf in #626
- INFO: avg_ttl from per-shard running sums, not a pass over every TTL key (#622) by @murtazayusuf in #630
- Expire a write's keys past their TTL first, and propagate no write that changed nothing (#628) by @murtazayusuf in #631
- CLIENT PAUSE WRITE waits out writes in flight and holds the ones that reach the gate after it (#629) by @murtazayusuf in #633
- Version 1.2.1-rc2 by @murtazayusuf in #635
- Benchmarks: 1.2.1 vector and memory/snapshot suites against 1.2.0 and competitors by @murtazayusuf in #638
- CI: run the Redis benchmark on Ubuntu 24.04 by @murtazayusuf in #646
- Version 1.2.1-rc3 by @murtazayusuf in #645
- Publish the package signing key's fingerprint by @murtazayusuf in #649
- Benchmarks 1.2.1: release checks on the final code by @murtazayusuf in #647
- Release 1.2.1 by @murtazayusuf in #648
New Contributors
- @matheussilvano made their first contribution in #325
- @ZYZ666-RGB made their first contribution in #480
- @paddor made their first contribution in #498
- @bfs2021 made their first contribution in #511
- @Ayege made their first contribution in #537
- @Anandb71 made their first contribution in #541
Full Changelog: v1.2.0...v1.2.1