Currently, once active defrag starts, we can not adjust
active_defrag_running
downwards. This is because active_defrag_running will be dynamically
compute
based on the fragmentation, we think we should not lower the effort when
the
fragmentation drops.
However, we need to note that active_defrag_running will also be
dynamically
computed based on configurations. In this case, we are not respecting
cycle-min
or cycle-max. Some people may realize halfway through that defrag
consumes a
lot and want to adjust it.
Previously we could only turn off activedefrag and then turn it on again
to
adjust active_defrag_running downwards. So in this PR, when a active
defrag
configuration change is made, we will re-compute it.
These configuration items are:
- active-defrag-cycle-min
- active-defrag-cycle-max
- active-defrag-threshold-upper
Since now a DB in cluster mode is divided into 16384 dicts, here
we directly check kvstoreDictSize instead of kvstoreSize, which
may have a higher probability that we can save the lookup.
The other change is a cleanup, obviously kvstoreGetHash should be
applied to the db->expires dicts.
When the dict is NULL, we also need to push resize_cursor, otherwise it
will keep doing useless continue here, and there is no way to resize the
other dict behind it.
Introduced in #12822.
---------
Co-authored-by: Oran Agra <oran@redislabs.com>
# Description
Gather most of the scattered `redisDb`-related code from the per-slot
dict PR (#11695) and turn it to a new data structure, `kvstore`. i.e.
it's a class that represents an array of dictionaries.
# Motivation
The main motivation is code cleanliness, the idea of using an array of
dictionaries is very well-suited to becoming a self-contained data
structure.
This allowed cleaning some ugly code, among others: loops that run twice
on the main dict and expires dict, and duplicate code for allocating and
releasing this data structure.
# Notes
1. This PR reverts the part of https://github.com/redis/redis/pull/12848
where the `rehashing` list is global (handling rehashing `dict`s is
under the responsibility of `kvstore`, and should not be managed by the
server)
2. This PR also replaces the type of `server.pubsubshard_channels` from
`dict**` to `kvstore` (original PR:
https://github.com/redis/redis/pull/12804). After that was done,
server.pubsub_channels was also chosen to be a `kvstore` (with only one
`dict`, which seems odd) just to make the code cleaner by making it the
same type as `server.pubsubshard_channels`, see
`pubsubtype.serverPubSubChannels`
3. the keys and expires kvstores are currenlty configured to allocate
the individual dicts only when the first key is added (unlike before, in
which they allocated them in advance), but they won't release them when
the last key is deleted.
Worth mentioning that due to the recent change the reply of DEBUG
HTSTATS changed, in case no keys were ever added to the db.
before:
```
127.0.0.1:6379> DEBUG htstats 9
[Dictionary HT]
Hash table 0 stats (main hash table):
No stats available for empty dictionaries
[Expires HT]
Hash table 0 stats (main hash table):
No stats available for empty dictionaries
```
after:
```
127.0.0.1:6379> DEBUG htstats 9
[Dictionary HT]
[Expires HT]
```
When db->expires_cursor==0, it means the DB is done the scanning,
we should exit the loop to avoid the useless scanning.
It is easy to see the active expire timeout in the modified test,
for example, let's assume that there is only 1 expired key in the
DB, and the size / buckets ratio is less than 1%, which means that
we will skip it in isExpiryDictValidForSamplingCb, and the return
value of expires_cursor is 0.
Because `data.sampled == 0` is always true, so `repeat` is also
always true, we will keep scanning the DB, but every time it is
skipped by the previous judgment (expires_cursor = 0), until the
timelimit is finally exhausted.
The JSON file lacks the following structural API changes:
- GEORADIUSBYMEMBER: add the ANY option for COUNT since 6.2.0.
- GEORADIUSBYMEMBER_RO: add the ANY option for COUNT since 6.2.0.
- GEORADIUS_RO: Added support for uppercase unit names since 7.0.0.
- GEORADIUSBYMEMBER_RO: Added support for uppercase unit names since
7.0.0.
---------
Signed-off-by: daz-3ux <daz-3ux@proton.me>
Co-authored-by: bodong.ybd <bodong.ybd@alibaba-inc.com>
Co-authored-by: Viktor Söderqvist <viktor.soderqvist@est.tech>
Co-authored-by: yangpengda.333 <yangpengda.333@bytedance.com>
Co-authored-by: Oran Agra <oran@redislabs.com>
Currently, We compute `db->avg_ttl` after each short `dbScan` sweep (a
few buckets without checking the time limit). But after each `dbScan`
sweep, we don't have much data and this makes the db->avg_ttl less
precise. For example, even if we scan the whole db, we can't get the
exact avg_ttl because we separate the data.
i.e. because of the running average, if we issue 16 calls to scan, we'll
give lower weight to the first one, and higher weight to the last one.
I think we should calculate `db->avg_ttl` until completing more of the
db iteration (judgement of time limit or the beginning of iterating next
db) because we have more sample data in this db and can get more
accurate result. In the best case, if we scan the whole db, we can get
the exact avg_ttl.
In this PR, we postpone the avg_ttl calculation until the judgement of
time limit or iteration of next db, so we can accumulate more data to
get more precise avg_ttl.
Note that we still need to make sure to decay the old TTLs at the same
speed as before, which is why we want to run the decay mechanism several
times, or use the Pow formula, see the comment in the code.
In my experiment, this PR can improve 89% or 52% accuracy in different
workload.
Co-authored-by: Oran Agra <oran@redislabs.com>
In Redis, rdb is produced in three scenarios mainly.
- backup, such as `bgsave` and `save` command
- full sync in replication
- aof rewrite if `aof-use-rdb-preamble` is yes
We also have some RDB flags to identify the purpose of rdb saving.
```C
/* flags on the purpose of rdb save or load */
#define RDBFLAGS_NONE 0 /* No special RDB loading. */
#define RDBFLAGS_AOF_PREAMBLE (1<<0) /* Load/save the RDB as AOF preamble. */
#define RDBFLAGS_REPLICATION (1<<1) /* Load/save for SYNC. */
```
But currently, it seems that these flags and purposes of rdb saving
don't exactly match. I find it in `rdbSaveRioWithEOFMark` which calls
`startSaving` with `RDBFLAGS_REPLICATION` but `rdbSaveRio` with
`RDBFLAGS_NONE`.
```C
int rdbSaveRioWithEOFMark(int req, rio *rdb, int *error, rdbSaveInfo *rsi) {
char eofmark[RDB_EOF_MARK_SIZE];
startSaving(RDBFLAGS_REPLICATION);
getRandomHexChars(eofmark,RDB_EOF_MARK_SIZE);
if (error) *error = 0;
if (rioWrite(rdb,"$EOF:",5) == 0) goto werr;
if (rioWrite(rdb,eofmark,RDB_EOF_MARK_SIZE) == 0) goto werr;
if (rioWrite(rdb,"\r\n",2) == 0) goto werr;
if (rdbSaveRio(req,rdb,error,RDBFLAGS_NONE,rsi) == C_ERR) goto werr;
if (rioWrite(rdb,eofmark,RDB_EOF_MARK_SIZE) == 0) goto werr;
stopSaving(1);
return C_OK;
werr: /* Write error. */
/* Set 'error' only if not already set by rdbSaveRio() call. */
if (error && *error == 0) *error = errno;
stopSaving(0);
return C_ERR;
}
```
In this PR, I refine the purpose of rdb saving with accurate flags.
Ci report this failure:
```
*** [err]: Don't rehash if used memory exceeds maxmemory after rehash in tests/unit/maxmemory.tcl
Expected '4098' to equal or match '4002'
WARNING: the new maxmemory value set via CONFIG SET (1176088) is smaller than the current memory usage (1231083)
```
It can be seen from the log that used_memory changed before we set
maxmemory.
The reason is that in #12819, in cron, in addition to trying to shrink,
we will
also tyring to expand. The dict was expanded by cron before we set
maxmemory,
causing the test to fail.
Before setting maxmemory, we only add 4095 keys to avoid triggering
resize.
When we use a timer to unblock a client in module, if the timer
period and the block timeout are very close, they will unblock the
client in the same event loop, and it will trigger the assertion.
The reason is that in moduleBlockedClientTimedOut we will protect
against re-processing, so we don't actually call updateStatsOnUnblock
(see #12817), so we are not able to reset the c->duration.
The reason is unblockClientOnTimeout() didn't realize that bc had
been unblocked. We add a function to the module to determine if bc
is blocked, and then use it in unblockClientOnTimeout() to exit.
There is the stack:
```
beforeSleep
blockedBeforeSleep
handleBlockedClientsTimeout
checkBlockedClientTimeout
unblockClientOnTimeout
unblockClient
resetClient
-- assertion, crash the server
'c->duration == 0' is not true
```
The block timeout is passed in the test case, but we do not pass
in the timeout_callback, and it will crash when unlocking. In this
case, in moduleBlockedClientTimedOut we will check timeout_callback.
There is the stack:
```
beforeSleep
blockedBeforeSleep
handleBlockedClientsTimeout
checkBlockedClientTimeout
unblockClientOnTimeout
replyToBlockedClientTimedOut
moduleBlockedClientTimedOut
-- timeout_callback is NULL, invalidFunctionWasCalled
bc->timeout_callback(&ctx,(void**)c->argv,c->argc);
```
Add a way to HSCAN a hash key, and get only the filed names.
Command syntax is now:
```
HSCAN key cursor [MATCH pattern] [COUNT count] [NOVALUES]
```
when `NOVALUES` is on, the command will only return keys in the hash.
---------
Co-authored-by: Viktor Söderqvist <viktor.soderqvist@est.tech>
Adds an ability to kill clients older than a specified age.
Also, fixed the age calculation in `catClientInfoString` to use
`commandTimeSnapshot`
instead of the old `server.unixtime`, and added missing documentation
for
`CLIENT KILL ID` to output of `CLIENT help`.
---------
Co-authored-by: Oran Agra <oran@redislabs.com>
tests consistently fail on timeout (sleep that's too short).
it now takes more time because in #12819 we iterate on all dicts, not
just non-empty ones.
it passed the PR's CI because it skips the `slow` tag, which might have
been misplaced, but now it is probably required.
with the fix, the tests take quite a lot of time:
```
[ok]: Redis can trigger resizing (1860 ms)
[ok]: Redis can rewind and trigger smaller slot resizing (744 ms)
```
before #12819:
```
[ok]: Redis can trigger resizing (309 ms)
[ok]: Redis can rewind and trigger smaller slot resizing (295 ms)
```
failure:
https://github.com/redis/redis/actions/runs/7704158180/job/20995931735
```
*** [err]: expire scan should skip dictionaries with lot's of empty buckets in tests/unit/expire.tcl
scan didn't handle slot skipping logic.
*** [err]: Redis can trigger resizing in tests/unit/other.tcl
Expected '[Dictionary HT]
Hash table 0 stats (main hash table):
table size: 128
number of elements: 5
[Expires HT]
Hash table 0 stats (main hash table):
No stats available for empty dictionaries
' to match '*table size: 8*' (context: type eval line 29 cmd {assert_match "*table size: 8*" [r debug HTSTATS 0]} proc ::test)
*** [err]: Redis can rewind and trigger smaller slot resizing in tests/unit/other.tcl
Expected '[Dictionary HT]
Hash table 0 stats (main hash table):
table size: 256
number of elements: 10
[Expires HT]
Hash table 0 stats (main hash table):
No stats available for empty dictionaries
' to match '*table size: 16*' (context: type eval line 27 cmd {assert_match "*table size: 16*" [r debug HTSTATS 0]} proc ::test)
```
This was introduced in #13004, missing this assignment.
It causes timeout to be a random value (may be less than now),
and then in `Unblock by timer` test, the client is unblocked
and then it call timeout_callback, since the callback is NULL,
the server will crash.
The crash stack is:
```
beforesleep
handleBlockedClientsTimeout
checkBlockedClientTimeout
unblockClientOnTimeout
replyToBlockedClientTimedOut
moduleBlockedClientTimedOut
-- the timeout_callback is NULL, invalidFunctionWasCalled
bc->timeout_callback(&ctx,(void**)c->argv,c->argc);
```
This allows specifying the timeout value for opening the TCP
connection to a server. The timeout, default 0 means no limit,
depending on the OS. It can be specified using the new `-t` switch.
revive #3764, fixes#3763
---------
Co-authored-by: Itamar Haber <itamar@redislabs.com>
Co-authored-by: yoav-steinberg <yoav@redislabs.com>
In #11012, we will reprocess command when client is unblocked on keys,
in some blocking commands, for example, in the XREADGROUP BLOCK
scenario,
because of the re-processing command, we will recalculate the block
timeout,
causing the blocking time to be reset.
This commit add a new CLIENT_REPROCESSING_COMMAND clent flag, explicitly
let the command know that it is being re-processed, later in
blockForKeys
we will not reset the timeout.
Affected BLOCK cases:
- list / zset / stream, added test cases for each.
Unaffected cases:
- module (never re-process the commands).
- WAIT / WAITAOF (never re-process the commands).
Fixes#12998.
The function `tryResizeHashTables` only attempts to shrink the dicts
that has keys (change from #11695), this was a serious problem until the
change in #12850 since it meant if all keys are deleted, we won't shrink
the dick.
But still, both dictShrink and dictExpand may be blocked by a fork child
process, therefore, the cron job needs to perform both dictShrink and
dictExpand, for not just non-empty dicts, but all dicts in DBs.
What this PR does:
1. Try to resize all dicts in DBs (not just non-empty ones, as it was
since #12850)
2. handle both shrink and expand (not just shrink, as it was since
forever)
3. Refactor some APIs about dict resizing (get rid of `htNeedsShrink`
`htNeedsShrink` `dictShrinkToFit`, and expose `dictShrinkIfNeeded`
`dictExpandIfNeeded` which already contains all the code of those
functions we get rid of, to make APIs more neat)
4. In the `Don't rehash if redis has child process` test, now that cron
would do resizing, we no longer need to write to DB after the child
process got killed, and can wait for the cron to expand the hash table.
Modules may want to handle allocation failures gracefully. Adding
RM_TryCalloc() and RM_TryRealloc() for it.
RM_TryAlloc() was added before:
https://github.com/redis/redis/pull/10541
We have not limited the value of maxmemory-samples in the past, it can
be set very large. If it is set very large, we will have stack overflow
in evictionPoolPopulate when we trigger the key eviction.
There is no reason for this config to be set too high, so just limit its
range to [1,64].
#### Problem Statement:
For any read/update operation during rehashing, we're doing ~10+ random
DRAM lookups to do the rehashing, as we are using the `rehashidx` to
rehash 10 buckets, whose dict entries most likely aren't cached in the
CPU or near the bucket we are operating on. If these random bucket are
empty, the rehashing process during that command execution is skipped.
#### Implementation:
For reducing the performance recession while dict is rehashing, we
determine the index at which the key would be stored in the 0th HT, we
check if that index has already been rehashed, if not we will rehash the
bucket containing the key and the bucket will be moved from 0th HT to
the 1st HT.
If the key has already been rehashed, we perform the random access
bucket rehash (using `rehashidx`) and we again verify if rehashing is
still ongoing and look up the key in the respective HT.
This ensures rehashing is not skipped in any command call and that we
rehash a particular bucket or random bucket in each call.
#### Changes in this PR:
- Added a new method `dictBucketRehash` to perform rehash on a single
bucket.
- Helper function `moveKeysInBucketOldtoNew` for `dictRehash` and
`dictBucketRehash` to move all the keys in a bucket from the old to the
new hash HT.
- Helper function `verifyMoreRehashRequired` for `dictRehash` and
`dictBucketRehash` to check if we have already rehashed the whole table
and if more rehashing is required.
### Benchmark:
- This PR still shows **~13%** improvement in the latency during
rehashing.
- Rehashing is now **~2%** faster for this PR when compared to unstable.
---------
Co-authored-by: Oran Agra <oran@redislabs.com>
Co-authored-by: Madelyn Olson <34459052+madolson@users.noreply.github.com>
Fix#9926 , and introduce an alternative method to prevent abuse of
transactions:
1. revert #5454 (which was blocking read-only transactions in OOM
state), and break the tie of MULTI state memory usage and the server OOM
state. Meaning that we'll limit the total memory a single client can
queue, and do that unconditionally regardless of the server being OOM or
not.
2. to prevent abuse of transactions, we use the
`client-query-buffer-limit` to restrict the size of the transaction.
Because the commands cached in the MULTI/EXEC queue have not been
executed yet, so they are also considered a part of the "query buffer"
in a broader sense. In other words, the commands in the MULTI queue and
the `querybuf` of the client together constitute the "query buffer".
When they exceed the limit, the connection will be disconnected.
The reasoning is that it's sensible to sends a single command with a
huge (1GB) argument, and it's sensible to sends a transaction with many
small commands, but it's probably not common to sends a long transaction
with many huge arguments (will consume a lot of memory before even being
executed).
If anyone runs into that, they can simply increase the
`client-query-buffer-limit` config.
P.S. To prevent DDoS attacks, unauthenticated clients have a separate
hard limit. Their query buffer should not exceed a maximum of 1MB. In
other words, if the query buffer of an unauthenticated client exceeds
1MB or the `client-query-buffer-limit` (if it is set to a value smaller
than 1MB,), the connection will be disconnected.
In the following case sender may be unknown, so we need to set up a
NULL check for sender:
```
/* If this is a MEET packet from an unknown node, we still process
* the gossip section here since we have to trust the sender because
* of the message type. */
if (!sender && type == CLUSTERMSG_TYPE_MEET)
clusterProcessGossipSection(hdr,link);
```
We have test cases for incr related commands with no key exist and
spaces in key and wrong type of key. However, we dont have test cases
covered for INCRBY INCRBYFLOAT DECRBY INCR DECR HINCRBY HINCRBYFLOAT
ZINCRBY with valid key and invalid value as argument, and float value to
incrby and decrby. So added test cases for the scenarios in incr.tcl.
Thank you!
In #11568 we removed the NOSCRIPT flag from commands, e.g. removing
NOSCRIPT flag from WAIT. Aiming to allow them in scripts and let them
implicitly behave in the non-blocking way.
This PR remove NOSCRIPT flag from WAITAOF just like WAIT (to be
symmetrical)).
And this PR also add BLOCKING flag for WAIT and WAITAOF.
This PR did some cleanups around function:
- drop the comment about Libraries Ctx, since we do have comment
in functionsLibCtx, no need to maintain multiple copies.
- remove outdated comment about the dropped Library description.
- remove unused desc and code vars in functionExtractLibMetaData.
- fix engines_nemory typo, changed it to engines_memory.
- remove outdated comment about FUNCTION CREATE and FUNCTION INFO,
FUNCTION CREATE was renamed to FUNCTION LOAD.
- Check in initServer whether the return of functionsInit is OK.
Currently slowlog gets disabled if slowlog-log-slower-than is set to less than zero. I think we should also disable it if slowlog-max-len is set to zero. We apply the same logic to acllog-max-len.
There have been occasional instances of memory corruption (though code bugs or bit flips) leading to invalid node information being gossiped around. To prevent this invalid information spreading, we verify the node IDs in received gossip are in an acceptable format, and disregard any gossiped nodes with invalid IDs. This PR uses the existing verifyClusterNodeId function to check the validity of the gossiped node IDs and if an invalid one is encountered, logs raw byte information to help debug the corruption.
---------
Co-authored-by: Madelyn Olson <madelyneolson@gmail.com>
background: some modules need to know the `dbid` information, such as
the function used during RDB loading:
```
robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
....
moduleInitIOContext(io,mt,rdb,&keyobj,dbid);
```
However, during replication, the "tempDb" created for diskless RDB
loading is not correctly set with the dbid. This leads to passing the
wrong dbid to the `rdbLoadObject` function (as tempDb uses zcalloc, all
ids are 0).
```
disklessLoadInitTempDb()->rdbLoadRioWithLoadingCtx()->
/* Read value */
val = rdbLoadObject(type,rdb,key,db->id,&error);
```
To fix it, set the correct ID (relative index) for the tempdb.
In #12838, we misuse the safe iterator of the client dict, so we can't
catch the synchronous release of the client if there is a bug.
Since we realize that clients (even subscribers) are released with async
free, we change the safe iterators of the client dict into unsafe
iterators in `pubsub.c`. And I also remove redundant code.
Before this change (most recently modified in
https://github.com/redis/redis/pull/12850#discussion_r1421406393), The
trigger for normal expand threshold was 100% utilization and the trigger
for normal shrink threshold was 10% (HASHTABLE_MIN_FILL).
While during fork (DICT_RESIZE_AVOID), when we want to avoid rehash, the
trigger thresholds were multiplied by 5 (`dict_force_resize_ratio`),
meaning 500% for expand and 2% (100/10/5) for shrink.
However, in `dictRehash` (the incremental rehashing), the rehashing
threshold for shrinking during fork (DICT_RESIZE_AVOID) was 20% by
mistake.
This meant that if a shrinking is triggered when `dict_can_resize` is
`DICT_RESIZE_ENABLE` which the threshold is 10%, the rehashing can
continue when `dict_can_resize` is `DICT_RESIZE_AVOID`.
This would cause unwanted CopyOnWrite damage.
It'll make sense to change the thresholds of the rehash trigger and the
thresholds of the incremental rehashing the same, however, in one we
compare the size of the hash table to the number of records, and in the
other we compare the size of ht[0] to the size of ht[1], so the formula
is not exactly the same.
to make things easier we change all the thresholds to powers of 2, so
the normal shrinking threshold is changed from 100/10 (i.e. 10%) to
100/8 (i.e. 12.5%), and we change the threshold during forks from 5 to
4, i.e. from 500% to 400% for expand, and from 2% (100/10/5) to 3.125%
(100/8/4)
Fix#12785 and other race condition issues.
See the following isolated comments.
The following report was obtained using SANITIZER thread.
```sh
make SANITIZER=thread
./runtest-moduleapi --config io-threads 4 --config io-threads-do-reads yes --accurate
```
1. Fixed thread-safe issue in RM_UnblockClient()
Related discussion:
https://github.com/redis/redis/pull/12817#issuecomment-1831181220
* When blocking a client in a module using `RM_BlockClientOnKeys()` or
`RM_BlockClientOnKeysWithFlags()`
with a timeout_callback, calling RM_UnblockClient() in module threads
can lead to race conditions
in `updateStatsOnUnblock()`.
- Introduced:
Version: 6.2
PR: #7491
- Touch:
`server.stat_numcommands`, `cmd->latency_histogram`, `server.slowlog`,
and `server.latency_events`
- Harm Level: High
Potentially corrupts the memory data of `cmd->latency_histogram`,
`server.slowlog`, and `server.latency_events`
- Solution:
Differentiate whether the call to moduleBlockedClientTimedOut() comes
from the module or the main thread.
Since we can't know if RM_UnblockClient() comes from module threads, we
always assume it does and
let `updateStatsOnUnblock()` asynchronously update the unblock status.
* When error reply is called in timeout_callback(), ctx is not
thread-safe, eventually lead to race conditions in `afterErrorReply`.
- Introduced:
Version: 6.2
PR: #8217
- Touch
`server.stat_total_error_replies`, `server.errors`,
- Harm Level: High
Potentially corrupts the memory data of `server.errors`
- Solution:
Make the ctx in `timeout_callback()` with `REDISMODULE_CTX_THREAD_SAFE`,
and asynchronously reply errors to the client.
2. Made RM_Reply*() family API thread-safe
Related discussion:
https://github.com/redis/redis/pull/12817#discussion_r1408707239
Call chain: `RM_Reply*()` -> `_addReplyToBufferOrList()` -> touch
server.current_client
- Introduced:
Version: 7.2.0
PR: #12326
- Harm Level: None
Since the module fake client won't have the `CLIENT_PUSHING` flag, even
if we touch server.current_client,
we can still exit after `c->flags & CLIENT_PUSHING`.
- Solution
Checking `c->flags & CLIENT_PUSHING` earlier.
3. Made freeClient() thread-safe
Fix#12785
- Introduced:
Version: 4.0
Commit:
3fcf959e60
- Harm Level: Moderate
* Trigger assertion
It happens when the module thread calls freeClient while the io-thread
is in progress,
which just triggers an assertion, and doesn't make any race condiaions.
* Touch `server.current_client`, `server.stat_clients_type_memory`, and
`clientMemUsageBucket->clients`.
It happens between the main thread and the module threads, may cause
data corruption.
1. Error reset `server.current_client` to NULL, but theoretically this
won't happen,
because the module has already reset `server.current_client` to old
value before entering freeClient.
2. corrupts `clientMemUsageBucket->clients` in
updateClientMemUsageAndBucket().
3. Causes server.stat_clients_type_memory memory statistics to be
inaccurate.
- Solution:
* No longer counts memory usage on fake clients, to avoid updating
`server.stat_clients_type_memory` in freeClient.
* No longer resetting `server.current_client` in unlinkClient, because
the fake client won't be evicted or disconnected in the mid of the
process.
* Judgment assertion `io_threads_op == IO_THREADS_OP_IDLE` only if c is
not a fake client.
4. Fixed free client args without GIL
Related discussion:
https://github.com/redis/redis/pull/12817#discussion_r1408706695
When freeing retained strings in the module thread (refcount decr), or
using them in some way (refcount incr), we should do so while holding
the GIL,
otherwise, they might be simultaneously freed while the main thread is
processing the unblock client state.
- Introduced:
Version: 6.2.0
PR: #8141
- Harm Level: Low
Trigger assertion or double free or memory leak.
- Solution:
Documenting that module API users need to ensure any access to these
retained strings is done with the GIL locked
5. Fix adding fake client to server.clients_pending_write
It will incorrectly log the memory usage for the fake client.
Related discussion:
https://github.com/redis/redis/pull/12817#issuecomment-1851899163
- Introduced:
Version: 4.0
Commit:
9b01b64430
- Harm Level: None
Only result in NOP
- Solution:
* Don't add fake client into server.clients_pending_write
* Add c->conn assertion for updateClientMemUsageAndBucket() and
updateClientMemoryUsage() to avoid same
issue in the future.
So now it will be the responsibility of the caller of both of them to
avoid passing in fake client.
6. Fix calling RM_BlockedClientMeasureTimeStart() and
RM_BlockedClientMeasureTimeEnd() without GIL
- Introduced:
Version: 6.2
PR: #7491
- Harm Level: Low
Causes inaccuracies in command latency histogram and slow logs, but does
not corrupt memory.
- Solution:
Module API users, if know that non-thread-safe APIs will be used in
multi-threading, need to take responsibility for protecting them with
their own locks instead of the GIL, as using the GIL is too expensive.
### Other issue
1. RM_Yield is not thread-safe, fixed via #12905.
### Summarize
1. Fix thread-safe issues for `RM_UnblockClient()`, `freeClient()` and
`RM_Yield`, potentially preventing memory corruption, data disorder, or
assertion.
2. Updated docs and module test to clarify module API users'
responsibility for locking non-thread-safe APIs in multi-threading, such
as RM_BlockedClientMeasureTimeStart/End(), RM_FreeString(),
RM_RetainString(), and RM_HoldString().
### About backpot to 7.2
1. The implement of (1) is not too satisfying, would like to get more
eyes.
2. (2), (3) can be safely for backport
3. (4), (6) just modifying the module tests and updating the
documentation, no need for a backpot.
4. (5) is harmless, no need for a backpot.
---------
Co-authored-by: Oran Agra <oran@redislabs.com>
When doing dict resizing, dictTypeResizeAllowed is used to judge whether
the new allocated memory for rehashing would cause OOM.
However when shrinking, we alloc `_dictNextExp(d->ht_used[0])` bytes of
memory, while in `dictTypeResizeAllowed` we still use
`_dictNextExp(d->ht_used[0]+1)` as the new allocated memory size. This
will overestimate the memory used by shrinking at special conditions,
causing a false OOM judgement.
We doing this in diskless on-empty-db mode, when diskless
loading fails, we will call emptyData to remove the half-loaded
data in case we started with an empty replica.
Now when a disk-based sync rdbLoad fails, we will call emptyData
too in case it loads partially incomplete data.
when the replica attempts another re-sync, it'll empty the dataset
again anyway, so this affects two things:
1. memory consumption in the time gap until the next rdb loading begins
2. if the unsynced replica is for some reason promoted, it would have kept
the partial dataset instead of being empty.
Before #12850, we will only try to shrink the dict in serverCron,
which we can control by using a child process, but now every time
we delete a key, the shrink check will be called.
In these test (added in #12802), we meant to disable the resizing,
but druing the delete, the dict will meet the force shrink, like
2 / 128 = 0.015 < 0.2, the delete will trigger a force resize and
will cause the test to fail.
In this commit, we try to keep the load factor at 3 / 128 = 0.023,
that is, do not meet the force shrink.
In the past we used integers to compare ratios, let us assume that
we have the following data in expanding:
```
used / size > 5
`80 / 16 > 5` is false
`81 / 16 > 5` is false
`95 / 16 > 5` is false
`96 / 16 > 5` is true
```
Because the integer result is rounded, our resize breaks the ratio
constraint, this has existed since the beginning, which resulted in
us not strictly following the ratio (shrink also has the same issue).
This PR change it to multiplication to avoid floating point
calculations.
The test have a race:
```
*** [err]: Redis can rewind and trigger smaller slot resizing in tests/unit/other.tcl
Expected '[Dictionary HT]
Hash table 0 stats (main hash table):
table size: 12
number of elements: 2
[Expires HT]
Hash table 0 stats (main hash table):
No stats available for empty dictionaries
' to match '*table size: 8*' (context: type eval line 12 cmd {assert_match "*table size: 8*" [r debug HTSTATS 0]} proc ::test)
```
When `r del "{alice}$j"` is executed in the loop, when the key is
deleted to [9, 12], the load factor has meet HASHTABLE_MIN_FILL,
if serverCron happens to trigger slot dict resize, then the test
will fail. Because there is not way to meet HASHTABLE_MIN_FILL in
the subsequent dels.
The solution is to avoid triggering the resize in advance. We can
use multi to delete them at once, or we can disable the resize.
Since we disabled resize in the previous test, the fix also uses
the method of disabling resize.
The test is introduced in #12802.
When we insert entries into dict, it may autonomously expand if needed.
However, when we delete entries from dict, it doesn't shrink to the
proper size. If there are few entries in a very large dict, it may cause
huge waste of memory and inefficiency when iterating.
The main keyspace dicts (keys and expires), are shrinked by cron
(`tryResizeHashTables` calls `htNeedsResize` and `dictResize`),
And some data structures such as zset and hash also do that (call
`htNeedsResize`) right after a loop of calls to `dictDelete`,
But many other dicts are completely missing that call (they can only
expand).
In this PR, we provide the ability to automatically shrink the dict when
deleting. The conditions triggering the shrinking is the same as
`htNeedsResize` used to have. i.e. we expand when we're over 100%
utilization, and shrink when we're below 10% utilization.
Additionally:
* Add `dictPauseAutoResize` so that flows that do mass deletions, will
only trigger shrinkage at the end.
* Rename `dictResize` to `dictShrinkToFit` (same logic as it used to
have, but better name describing it)
* Rename `_dictExpand` to `_dictResize` (same logic as it used to have,
but better name describing it)
related to discussion
https://github.com/redis/redis/pull/12819#discussion_r1409293878
---------
Co-authored-by: Oran Agra <oran@redislabs.com>
Co-authored-by: zhaozhao.zz <zhaozhao.zz@alibaba-inc.com>
Regarding how to obtain the hash slot of a key, there is an optimization
in `getKeySlot()`, it is used to avoid redundant hash calculations for
keys: when the current client is in the process of executing a command,
it can directly use the slot of the current client because the slot to
access has already been calculated in advance in `processCommand()`.
However, scripts are a special case where, in default mode or with
`allow-cross-slot-keys` enabled, they are allowed to access keys beyond
the pre-declared range. This means that the keys they operate on may not
belong to the slot of the pre-declared keys. Currently, when the
commands in a script are executed, the slot of the original client
(i.e., the current client) is not correctly updated, leading to
subsequent access to the wrong slot.
This PR fixes the above issue. When checking the cluster constraints in
a script, the slot to be accessed by the current command is set for the
original client (i.e., the current client). This ensures that
`getKeySlot()` gets the correct slot cache.
Additionally, the following modifications are made:
1. The 'sort' and 'sort_ro' commands use `getKeySlot()` instead of
`c->slot` because the client could be an engine client in a script and
can lead to potential bug.
2. `getKeySlot()` is also used in pubsub to obtain the slot for the
channel, standardizing the way slots are retrieved.
Change the calculation method of bytes_per_key to make it closer to
the true average key size. The calculation method is as follows:
mh->bytes_per_key = mh->total_keys ? (mh->dataset / mh->total_keys) : 0;
The test was introduced in #10745, but we forgot to add it to the
test_helper.tcl, so our CI did not actually run it. This PR adds it
and ensures it passes CI tests.
