- 14 Aug, 2024 1 commit
-
-
debing.sun authored
Fixed a missing from #13117. When the number of streams is incorrect, the error message for `XREAD` needs to include the '+' symbol.
-
- 11 Aug, 2024 1 commit
-
-
Moti Cohen authored
Hash field expiration is optimized to avoid frequent update global HFE DS for each field deletion. Eventually active-expiration will run and update or remove the hash from global HFE DS gracefully. Nevertheless, statistic "subexpiry" might reflect wrong number of hashes with HFE to the user if HDEL deletes the last field with expiration in hash (yet there are more fields without expiration). Following this change, if HDEL the last field with expiration in the hash then take care to remove the hash from global HFE DS as well.
-
- 08 Aug, 2024 2 commits
-
-
LuMingYinDetect authored
Fix memory leak related to variable slot_nodes in the clusterManagerFixSlotsCoverage() function. --------- Co-authored-by:debing.sun <debing.sun@redis.com>
-
debing.sun authored
This PR is based on the commits from PR https://github.com/valkey-io/valkey/pull/52. Ref: https://github.com/redis/redis/pull/12760 Close https://github.com/redis/redis/issues/13401 This PR will replace https://github.com/redis/redis/pull/13449 Fixes compatibilty of Redis cluster (7.2 - extensions enabled by default) with older Redis cluster (< 7.0 - extensions not handled) . With some of the extensions enabled by default in 7.2 version, new nodes running 7.2 and above start sending out larger clusterbus message payload including the ping extensions. This caused an incompatibility with node running engine versions < 7.0. Old nodes (< 7.0) would receive the payload from new nodes (> 7.2) would observe a payload length (totlen) > (estlen) and would perform an early exit and won't process the message. This fix does the following things: 1. Always set `CLUSTERMSG_FLAG0_EXT_DATA`, because during the meet phase, we do not know whether the connected node supports ext data, we need to make sure that it knows and send back its ext data if it has. 2. If another node does not support ext data, we will not send it ext data to avoid the handshake failure due to the incorrect payload length. Note: A successful `PING`/`PONG` is required as a sender for a given node to be marked as `CLUSTERMSG_FLAG0_EXT_DATA` and then extensions message will be sent to it. This could cause a slight delay in receiving the extensions message(s). --------- Signed-off-by:
Harkrishn Patro <harkrisp@amazon.com> Co-authored-by:
Harkrishn Patro <harkrisp@amazon.com> --------- Signed-off-by:
Harkrishn Patro <harkrisp@amazon.com> Co-authored-by:
Harkrishn Patro <harkrisp@amazon.com>
-
- 06 Aug, 2024 1 commit
-
-
debing.sun authored
First, we need to ensure that `curmaster` in `clusterUpdateSlotsConfigWith()` is not NULL in the line https://github.com/redis/redis/blob/82f00f5179720c8cee6cd650763d184ba943be92/src/cluster_legacy.c#L2320 otherwise, it will crash in the https://github.com/redis/redis/blob/82f00f5179720c8cee6cd650763d184ba943be92/src/cluster_legacy.c#L2395 So when loading cluster node config, we need to ensure that the following conditions are met: 1. A node must be at least one of the master or replica. 2. If a node is a replica, its master can't be NULL.
-
- 05 Aug, 2024 3 commits
-
-
Josh Hershberg authored
This and the previous commit make the cluster shards command a generic implementation instead of a specific implementation for each cluster API implementation. This commit (a) adds functions to the cluster API and (b) modifies the cluster shards cmd implementation to use cluster API functions instead of directly accessing the legacy clustering implementation. Signed-off-by:Josh Hershberg <yehoshua@redis.com>
-
Josh Hershberg authored
This and the next following commit makes the cluster shards command a generic implementation instead of a specific implementation for each cluster API implementation. This commit simply moves the cluster shards implementation from cluster_legacy.c to cluster.c without changing the implementation at all. The reason for doing so was to help with reviewing the changes in the diff. Signed-off-by:Josh Hershberg <yehoshua@redis.com>
-
Zhongxian Pan authored
## Replace bit shift with `__builtin_ctzll` in HyperLogLog Builtin function `__builtin_ctzll` is more effective than bit shift even though "in the average case there are high probabilities to find a 1 after a few iterations" mentioned in the source file comment. --------- Co-authored-by:debing.sun <debing.sun@redis.com>
-
- 04 Aug, 2024 1 commit
-
-
Moti Cohen authored
H[P]TTL should be marked as NONDETERMINISTIC_OUTPUT just like [P]TTL.
-
- 03 Aug, 2024 1 commit
-
-
Vitah Lin authored
When the server restarts while the CLI is connecting, the reconnection does not automatically select the previous db. This may lead users to believe they are still in the previous db, in fact, they are in db0. This PR will automatically reset the current dbnum and `cliSelect()` again when reconnecting. --------- Co-authored-by:debing.sun <debing.sun@redis.com>
-
- 01 Aug, 2024 2 commits
-
-
c8ef authored
-
debing.sun authored
Close https://github.com/redis/redis/issues/13414 When the cluster's master node fails and is switched to another node, the first node in the shard node list (the old master) is no longer valid. Add a new method clusterGetMasterFromShard() to obtain the current master.
-
- 30 Jul, 2024 1 commit
-
-
debing.sun authored
Fix #13337 Ths PR fixes fixed two bugs that caused lag calculation errors. 1. When the latest tombstone is before the first entry, the tombstone may stil be after the last id of consume group. 2. When a tombstone is after the last id of consume group, the group's counter will be invalid, we should caculate the entries_read by using estimates.
-
- 28 Jul, 2024 1 commit
-
-
Lior Kogan authored
-
- 25 Jul, 2024 1 commit
-
-
Moti Cohen authored
-
- 24 Jul, 2024 1 commit
-
-
Moti Cohen authored
Modify RDB_TYPE_HASH_METADATA layout to store expiration times relative to the minimum expiration time, which is written at the start as absolute time.
-
- 22 Jul, 2024 1 commit
-
-
Oran Agra authored
Recently in #13361, i attempted to fix a race between FLUSHALL and BGSAVE, where despite calling killRDBChild, the backgroundSaveDoneHandler will terminate with success. Turns out that even if the child didn't yet exit, there's a chance it'll still miss our signal and exit with success. in that case, we will still mess up the dirty counter (deducting dirty_before_bgsave) which is reset by FLUSHALL, and override the synchronous rdb file we saved. instead, we'll set a flag to treat the next done handler as a failed one.
-
- 16 Jul, 2024 2 commits
-
-
debing.sun authored
Nowdays we do not trigger LUA GC after loading lua script. This means that when a large number of scripts are loaded, such as when functions are propagating from the master to the replica, if the LUA scripts are never touched on the replica, the garbage might remain there indefinitely. Before this PR, we would share a gc_count between scripts and functions. This means that, under certain circumstances, the GC trigger for scripts and functions was not fair. For example, loading a large number of scripts followed by a small number of functions could result in the functions triggering GC. In this PR, we assign a unique `gc_count` to each of them, so the GC triggers between them will no longer affect each other. on the other hand, this PR will to bring regession for script loading commands(`FUNCTION LOAD` and `SCRIPT LOAD`), but they are not hot path, we can ignore it, and it will be replaced https://github.com/redis/redis/pull/13375 in the future. --------- Co-authored-by:
Oran Agra <oran@redislabs.com>
-
debing.sun authored
When we terminate the diskless RDB saving child process and, at the same time, we start a new BGSAVE for new replicas, we should not delete the RDB read event. Otherwise, these replicas will never receive a response. this is a result of the recent change in https://github.com/redis/redis/pull/13361 --------- Co-authored-by:
oranagra <oran@redislabs.com>
-
- 15 Jul, 2024 1 commit
-
-
debing.sun authored
This PR is based on the commits from PR https://github.com/valkey-io/valkey/pull/670. ## Description While exploring hotspots with profiling some benchmark workloads, we noticed the high cycles ratio of `prepareClientToWrite`, taking about 9% of the CPU of `smembers`, `lrange` commands. After deep dive the code logic, we thought we can gain the performance by reducing the redundant call of `prepareClientToWrite` when call addReply* continuously. For example: In https://github.com/valkey-io/valkey/blob/unstable/src/networking.c#L1080-L1082 , `prepareClientToWrite` is called three times in a row. --------- Signed-off-by:
Lipeng Zhu <lipeng.zhu@intel.com> Co-authored-by:
Lipeng Zhu <lipeng.zhu@intel.com> Co-authored-by:
Wangyang Guo <wangyang.guo@intel.com>
-
- 14 Jul, 2024 1 commit
-
-
guybe7 authored
128 is not enough chars when we're talking about commands like RESTORE. Of course, it's impossible to find the perfect number, but 1024 is better than 128, and it's not obscenely large.
-
- 11 Jul, 2024 2 commits
-
-
guybe7 authored
To be more similar to EXPIRE-like commands, which emit a "del" notification if the expire-time is in the past
-
guybe7 authored
This commit reverts the deletion of the condition `!bc->blocked_on_keys` that was accidentally introduced by https://github.com/redis/redis/pull/12817 . In case a blocked-on-keys module client is unblocked both `moduleUnblockClientOnKey` and `moduleHandleBlockedClients` are called which resulted in `updateStatsOnUnblock` being called twice Now, that `moduleHandleBlockedClients` doesn't call `updateStatsOnUnblock` in case of unblocked module key-blocked clients, in the unlikely event that the module decides to call `RM_UnblockClient` on a key-blocked client, we need to call `updateStatsOnUnblock` from within `moduleBlockedClientTimedOut`, but since `moduleBlockedClientTimedOut` is not tread-safe we can't call it directly from withing `RM_UnblockClient`. Added a new flag `blocked_on_keys_explicit_unblock` for that specific case, which will cause `moduleBlockedClientTimedOut` to be called from `moduleHandleBlockedClients` (which is only called from the main thread) --------- Co-authored-by:
debing.sun <debing.sun@redis.com>
-
- 10 Jul, 2024 1 commit
-
-
debing.sun authored
### Issue The current implementation of `FUNCTION FLUSH` command uses `lua_unref()` to unreference script closures in Lua vm. However, invoking `lua_unref()` during lazy free (`ASYNC` argument) is risky since it is not thread-safe. Another issue is that using `lua_unref()` to unreference references does not trigger GC, This can result in the Lua VM leaves a significant amount of garbage, which may never be cleaned up if not properly GC. ### Solution The proposed solution is to completely rebuild the engines, resulting in a brand new Lua VM. --------- Co-authored-by:meir <meir@redis.com>
-
- 09 Jul, 2024 1 commit
-
-
debing.sun authored
This PR is based on the commits from PR #11747. In the event of an assertion failure, hide command arguments from the operator. In some cases, private client information can be voluntarily exposed when a redis instance crashes due to an assertion failure. This commit prevent וnintentional client info exposure. Operators can still access the hidden data, but they must actively request it. Any of the client info commands remains the unchanged. ### Config Add a new config `hide-user-data-from-log` to turn this feature on and off, default off. --------- Co-authored-by:
naglera <anagler123@gmail.com> Co-authored-by:
naglera <58042354+naglera@users.noreply.github.com>
-
- 07 Jul, 2024 1 commit
-
-
Moti Cohen authored
* Following this feature, Redis (ROF) may implement flow that allows hashes to be dumped directly from RDB to FLUSH without parsing. In this scenario, it is still essential to determine when to update hashes due to expired fields. By writing and reading the next minimum hash-field expiration before serializing objects to and from RDB, we can effectively track and expire hash fields without the need to parse the hash during loading. Before: #define RDB_TYPE_HASH_METADATA 22 #define RDB_TYPE_HASH_LISTPACK_EX 23 After: /* Hash with HFEs. Doesn't attach min TTL at start */ #define RDB_TYPE_HASH_METADATA_PRE_GA 22 /* Hash LP with HFEs. Doesn't attach min TTL at start */ #define RDB_TYPE_HASH_LISTPACK_EX_PRE_GA 23 /* Hash with HFEs. Attach min TTL at start */ #define RDB_TYPE_HASH_METADATA 24 /* Hash LP with HFEs. Attach min TTL at start */ #define RDB_TYPE_HASH_LISTPACK_EX 25 * Manually test loading RDB file before the change and verify hash and its HFEs is as expected. * Added `subexpires` counter to `redis-check-rdb`
-
- 04 Jul, 2024 1 commit
-
-
debing.sun authored
1. Add help for `DEBUG SCRIPT` command. 2. Remove a duplicate `getLuaScripts()` which is same as `evalScriptsDict()`.
-
- 03 Jul, 2024 1 commit
-
-
Filipe Oliveira (Redis) authored
The following PR does the following changes based upon on CPU profile info. The `getNodeByQuery` function represents 8.2% of an overhead of 12.3% when comparing single shard cluster with standalone. Proposed changes: - inlinging keyHashSlot to reduce overhead of that function call - Reduce duplicate calls to getCommandFlags within getNodeByQuery The above changes represent an improvement of approximately 5% on the achievable ops/sec. Co-authored-by:filipecosta90 <filipecosta.90@gmail.com>
-
- 02 Jul, 2024 1 commit
-
-
Moti Cohen authored
* INFO command : rename `hashes_with_expiry_fields` to `subexpiry` * INFO command : rename `expired_hash_fields` to `expired_subkeys` * Fix statistic of `expired_subkeys` to count also lazy expired * Remove TODOs comments leftover in TCL * Fix potential flaky test of rdb load of hash-field-expiration
-
- 01 Jul, 2024 2 commits
-
-
Oran Agra authored
If we run FLUSHALL when the 'save' config is set, and there's a fork child ding BGSAVE, there's a chance the child is already finished, and the parent process is unaware of it. in that case the child will not get the kill signal and will finish successfully, but the parent process thinks it killed it and will reset the dirty counter to 0, then the backgroundSaveDoneHandlerDisk method can set the dirty counter to a negative value.
-
Oran Agra authored
getKeysUsingKeySpece had the range check AFTER the allocation, of the keys buffer, which could lead to an OOM panic when invalid arguments are provided, leading to an overflow. The allocated memory is only used after the range check, so there's no risk of buffer overrun. The OOM panic can happen on 32bit builds, or 64 builds running on systems with less than 4GB of RAM, and is reachable via the COMMAND GETKEYSANDFLAGS, and ACL key name validation.
-
- 26 Jun, 2024 2 commits
-
-
Moti Cohen authored
There was wrong preliminary assumption that we can optionally provide vector of arguments more than count. This is error-prone approach that leaded to actual error in that case. This PR enforce that vector of argument match count. Also fixed flaky HRANDFIELD test.
-
debing.sun authored
In certain situations, we might generate a large number of propagates (e.g., multi/exec, Lua script, or a single command generating tons of propagations) within an event loop. During the process of propagating to a replica, if the replica is disconnected(marked as CLIENT_CLOSE_ASAP) due to exceeding the output buffer limit, we should remove its reference to the global replication buffer to avoid the global replication buffer being unable to be properly trimmed due to being referenced. --------- Co-authored-by:oranagra <oran@redislabs.com>
-
- 25 Jun, 2024 1 commit
-
-
Moti Cohen authored
H(P)EXPIREAT command might delete fields in case the absolute time is in the past. Those HDELs need to be propagated as well. In general, as we need to propagate H(P)EXPIRE(AT) command to the replica, each field that is mentioned in the command should be categorized into one of the four options: 1. Managed to update field’s expiration time - propagate it to replica as part of the HPEXPIREAT command. 2. Deleted the field because the time is in the past - propagate also HDEL command to delete the field and remove the field from the propagated HPEXPIREAT. 3. Condition not met for the field - Remove the field from the propagated HPEXPIREAT command. 4. Field does not exists - Remove the field from the propagated HPEXPIREAT command. If none of the provided fields match option number 1, then avoid also propagating the HPEXPIREAT command to the replica. This approach is aligned with the EXPIRE command: If a given key has already expired, then DEL will be propagated instead of EXPIRE command. If condition not met, then command will be rejected. Otherwise, EXPIRE command will be propagated for given key.
-
- 24 Jun, 2024 1 commit
-
-
Moti Cohen authored
Considerations for the selected imp of HRANDFIELD & HFE feature: HRANDFIELD might access any of the fields in the hash as some of them might be expired. And so the Implementation of HRANDFIELD along with HFEs might be one of the two options: 1. Expire hash-fields before diving into handling HRANDFIELD. 2. Refine HRANDFIELD cases to deal with expired fields. Regarding the first option, as reference, the command RANDOMKEY also declareson O(1) complexity, yet might be stuck on a very long (but not infinite) loop trying to find non-expired keys. Furthermore RANDOMKEY also evicts expired keys along the way even though it is categorized as a read-only command. Note that the case of HRANDFIELD is more lightweight versus RANDOMKEY since HFEs have much more effective and aggressive active-expiration for fields behind. The second option introduces additional implementation complexity to HRANDFIELD. We could further refine HRANDFIELD cases to differentiate between scenarios with many expired fields versus few expired fields, and adjust based on the percentage of expired fields. However, this approach could still lead to long loops or necessitate expiring fields before selecting them. For the “lightweight” cases it is also expected to have a lightweight expiration. Considering the pros and cons, and the fact that HRANDFIELD is an infrequent command (particularly with HFEs) and the fact we have effective active-expiration behind for hash-fields, it is better to keep it simple and choose option number 1. Other changes: * Don't mark command dirty by internal hashTypeExpire(). It causes to read only command of HRANDFIELD to be accidently propagated (This flag should be indicated at higher level, by the command functions). * Align `hashTypeExpireIfNeeded()` and `hashTypeGetValue()` to be more aligned with `expireIfNeeded()` logic of keyspace.
-
- 21 Jun, 2024 2 commits
-
-
AcherTT authored
Add two new debug commands for outputing script. 1. `DEBUG SCRIPT LIST` Output all scripts. 2. `DEBUG SCRIPT <sha1>` Output a specific script. Close #3846 -
debing.sun authored
Related to #12647 1. Make clear that `RM_Replicate` and `RM_ReplicateVerbatim` are non-thread safe. 2. Make clear that `RM_Replicate` and `RM_ReplicateVerbatim` are alwarys wrapped into MULTI in any case.
-
- 20 Jun, 2024 1 commit
-
-
Moti Cohen authored
As part of HFE feature, the logic of rdbLoadObject() was wrongly modified to indicate of loaded empty hash from RDB as hash that all its fields got expired. Rollback to `emptykey` logic. This function should load blindly all fields, expired or not. Manually verified. Few more minor fixes: - remove hash double check of emptyKey - Fix from `sds` to `hfield` in rdbLoadObject() (not really a bug. Both are of type char*) - Revert code rdbLoadObject() to get dbid instead of db
-
- 18 Jun, 2024 1 commit
-
-
Filipe Oliveira (Redis) authored
reduce getNodeByQuery CPU time by using less cache lines (from 2064 Bytes struct to 64 Bytes): reduces LLC misses and Memory Loads (#13296) The following PR goes from 33 cacheline on getKeysResult struct (by default has 256 static buffer) ``` root@hpe10:~/redis# pahole -p ./src/server.o -C getKeysResult typedef struct { keyReference keysbuf[256]; /* 0 2048 */ /* --- cacheline 32 boundary (2048 bytes) --- */ /* typedef keyReference */ struct { int pos; int flags; } *keys; /* 2048 8 */ int numkeys; /* 2056 4 */ int size; /* 2060 4 */ /* size: 2064, cachelines: 33, members: 4 */ /* last cacheline: 16 bytes */ } getKeysResult; ``` to 1 cacheline with a static buffer of 6 keys per command): ``` root@hpe10:~/redis# pahole -p ./src/server.o -C getKeysResult typedef struct { int numkeys; /* 0 4 */ int size; /* 4 4 */ keyReference keysbuf[6]; /* 8 48 */ /* typedef keyReference */ struct { int pos; int flags; } *keys; /* 56 8 */ /* size: 64, cachelines: 1, members: 4 */ } getKeysResult; ``` we get around 1.5% higher ops/sec, and a confirmation of around 15% less LLC loads on getNodeByQuery and 37% less Stores. Function / Call Stack | CPU Time: Difference | CPU Time: 9462436fa444e746716845b1d807c74d8945831b | CPU Time: this PR | Loads: Difference | Loads: 9462436fa444e746716845b1d807c74d8945831b | Loads: this PR | Stores: Difference | Stores: 9462436fa444e746716845b1d807c74d8945831b | Stores: This PR -- | -- | -- | -- | -- | -- | -- | -- | -- | -- getNodeByQuery | 0.753767 | 1.57118 | 0.817416 | 144297829 (15% less loads) | 920575969 | 776278140 | 367607824 (37% less stores) | 991642384 | 624034560 ## results on client side ### baseline ``` taskset -c 2,3 memtier_benchmark -s 192.168.1.200 --port 6379 --authenticate perf --cluster-mode --pipeline 10 --data-size 100 --ratio 1:0 --key-pattern P:P --key-minimum=1 --key-maximum 1000000 --test-time 180 -c 25 -t 2 --hide-histogram Writing results to stdout [RUN #1] Preparing benchmark client... [RUN #1] Launching threads now... [RUN #1 100%, 180 secs] 0 threads: 110333450 ops, 604992 (avg: 612942) ops/sec, 84.75MB/sec (avg: 85.86MB/sec), 0.82 (avg: 0.81) msec latency 2 Threads 25 Connections per thread 180 Seconds ALL STATS ====================================================================================================================================================== Type Ops/sec Hits/sec Misses/sec MOVED/sec ASK/sec Avg. Latency p50 Latency p99 Latency p99.9 Latency KB/sec ------------------------------------------------------------------------------------------------------------------------------------------------------ Sets 612942.14 --- --- 0.00 0.00 0.81332 0.80700 1.26300 2.92700 87924.12 Gets 0.00 0.00 0.00 0.00 0.00 --- --- --- --- 0.00 Waits 0.00 --- --- --- --- --- --- --- --- --- Totals 612942.14 0.00 0.00 0.00 0.00 0.81332 0.80700 1.26300 2.92700 87924.12 ``` ### comparison ``` taskset -c 2,3 memtier_benchmark -s 192.168.1.200 --port 6379 --authenticate perf --cluster-mode --pipeline 10 --data-size 100 --ratio 1:0 --key-pattern P:P --key-minimum=1 --key-maximum 1000000 --test-time 180 -c 25 -t 2 --hide-histogram Writing results to stdout [RUN #1] Preparing benchmark client... [RUN #1] Launching threads now... [RUN #1 100%, 180 secs] 0 threads: 111731310 ops, 610195 (avg: 620707) ops/sec, 85.48MB/sec (avg: 86.95MB/sec), 0.82 (avg: 0.80) msec latency 2 Threads 25 Connections per thread 180 Seconds ALL STATS ====================================================================================================================================================== Type Ops/sec Hits/sec Misses/sec MOVED/sec ASK/sec Avg. Latency p50 Latency p99 Latency p99.9 Latency KB/sec ------------------------------------------------------------------------------------------------------------------------------------------------------ Sets 620707.72 --- --- 0.00 0.00 0.80312 0.79900 1.23900 2.87900 89037.78 Gets 0.00 0.00 0.00 0.00 0.00 --- --- --- --- 0.00 Waits 0.00 --- --- --- --- --- --- --- --- --- Totals 620707.72 0.00 0.00 0.00 0.00 0.80312 0.79900 1.23900 2.87900 89037.78 ``` Co-authored-by:filipecosta90 <filipecosta.90@gmail.com>
-
- 14 Jun, 2024 1 commit
-
-
Ozan Tezcan authored
Currently, HFE commands reply with empty array if the key does not exist. Though, non-existing key and empty key is the same thing. It means fields given in the command do not exist in the empty key. So, replying with an array of 'no field' error codes (-2) suits better to Redis logic. e.g. Similarly, `hmget` returns array of nulls if the key does not exist. After this PR: ``` 127.0.0.1:6379> hpersist missingkey fields 2 a b 1) (integer) -2 2) (integer) -2 ```
-