Implement Multi-Part AOF mechanism to avoid overheads during AOFRW.
Introducing a folder with multiple AOF files tracked by a manifest file.
The main issues with the the original AOFRW mechanism are:
* buffering of commands that are processed during rewrite (consuming a lot of RAM)
* freezes of the main process when the AOFRW completes to drain the remaining part of the buffer and fsync it.
* double disk IO for the data that arrives during AOFRW (had to be written to both the old and new AOF files)
The main modifications of this PR:
1. Remove the AOF rewrite buffer and related code.
2. Divide the AOF into multiple files, they are classified as two types, one is the the `BASE` type,
it represents the full amount of data (Maybe AOF or RDB format) after each AOFRW, there is only
one `BASE` file at most. The second is `INCR` type, may have more than one. They represent the
incremental commands since the last AOFRW.
3. Use a AOF manifest file to record and manage these AOF files mentioned above.
4. The original configuration of `appendfilename` will be the base part of the new file name, for example:
`appendonly.aof.1.base.rdb` and `appendonly.aof.2.incr.aof`
5. Add manifest-related TCL tests, and modified some existing tests that depend on the `appendfilename`
6. Remove the `aof_rewrite_buffer_length` field in info.
7. Add `aof-disable-auto-gc` configuration. By default we're automatically deleting HISTORY type AOFs.
It also gives users the opportunity to preserve the history AOFs. just for testing use now.
8. Add AOFRW limiting measure. When the AOFRW failures reaches the threshold (3 times now),
we will delay the execution of the next AOFRW by 1 minute. If the next AOFRW also fails, it will be
delayed by 2 minutes. The next is 4, 8, 16, the maximum delay is 60 minutes (1 hour). During the limit
period, we can still use the 'bgrewriteaof' command to execute AOFRW immediately.
9. Support upgrade (load) data from old version redis.
10. Add `appenddirname` configuration, as the directory name of the append only files. All AOF files and
manifest file will be placed in this directory.
11. Only the last AOF file (BASE or INCR) can be truncated. Otherwise redis will exit even if
`aof-load-truncated` is enabled.
Co-authored-by: Oran Agra <oran@redislabs.com>
This commit adds DUMP RESTORES tests for the -x and -X options.
I wanted to add it in #9980 which introduce the -X option, but
back then i failed due to some errors (related to redis-cli call).
To avoid data loss, this commit adds a grace period for lagging replicas to
catch up the replication offset.
Done:
* Wait for replicas when shutdown is triggered by SIGTERM and SIGINT.
* Wait for replicas when shutdown is triggered by the SHUTDOWN command. A new
blocked client type BLOCKED_SHUTDOWN is introduced, allowing multiple clients
to call SHUTDOWN in parallel.
Note that they don't expect a response unless an error happens and shutdown is aborted.
* Log warning for each replica lagging behind when finishing shutdown.
* CLIENT_PAUSE_WRITE while waiting for replicas.
* Configurable grace period 'shutdown-timeout' in seconds (default 10).
* New flags for the SHUTDOWN command:
- NOW disables the grace period for lagging replicas.
- FORCE ignores errors writing the RDB or AOF files which would normally
prevent a shutdown.
- ABORT cancels ongoing shutdown. Can't be combined with other flags.
* New field in the output of the INFO command: 'shutdown_in_milliseconds'. The
value is the remaining maximum time to wait for lagging replicas before
finishing the shutdown. This field is present in the Server section **only**
during shutdown.
Not directly related:
* When shutting down, if there is an AOF saving child, it is killed **even** if AOF
is disabled. This can happen if BGREWRITEAOF is used when AOF is off.
* Client pause now has end time and type (WRITE or ALL) per purpose. The
different pause purposes are *CLIENT PAUSE command*, *failover* and
*shutdown*. If clients are unpaused for one purpose, it doesn't affect client
pause for other purposes. For example, the CLIENT UNPAUSE command doesn't
affect client pause initiated by the failover or shutdown procedures. A completed
failover or a failed shutdown doesn't unpause clients paused by the CLIENT
PAUSE command.
Notes:
* DEBUG RESTART doesn't wait for replicas.
* We already have a warning logged when a replica disconnects. This means that
if any replica connection is lost during the shutdown, it is either logged as
disconnected or as lagging at the time of exit.
Co-authored-by: Oran Agra <oran@redislabs.com>
This is needed in order to ease the deployment of functions for ephemeral cases, where user
needs to spin up a server with functions pre-loaded.
#### Details:
* Added `--functions-rdb` option to _redis-cli_.
* Functions only rdb via `REPLCONF rdb-filter-only functions`. This is a placeholder for a space
separated inclusion filter for the RDB. In the future can be `REPLCONF rdb-filter-only
"functions db:3 key-patten:user*"` and a complementing `rdb-filter-exclude` `REPLCONF`
can also be added.
* Handle "slave requirements" specification to RDB saving code so we can use the same RDB
when different slaves express the same requirements (like functions-only) and not share the
RDB when their requirements differ. This is currently just a flags `int`, but can be extended to
a more complex structure with various filter fields.
* make sure to support filters only in diskless replication mode (not to override the persistence file),
we do that by forcing diskless (even if disabled by config)
other changes:
* some refactoring in rdb.c (extract portion of a big function to a sub-function)
* rdb_key_save_delay used in AOFRW too
* sendChildInfo takes the number of updated keys (incremental, rather than absolute)
Co-authored-by: Oran Agra <oran@redislabs.com>
This pr is mainly to solve the problem that redis process cannot be exited normally, due to changes in #10003.
When a test uses the `key-load-delay` config to delay loading, but does not reset it at the end of the test, will lead to server wait for the loading to reach the event
loop (once in 2mb) before actually shutting down.
There are two changes in this commit:
1. Add -X option to redis-cli.
Currently `-x` can only be used to provide the last argument,
so you can do `redis-cli dump keyname > key.dump`,
and then do `redis-cli -x restore keyname 0 < key.dump`.
But what if you want to add the replace argument (which comes last?).
oran suggested adding such usage:
`redis-cli -X <tag> restore keyname <tag> replace < key.dump`
i.e. you're able to provide a string in the arguments that's gonna be
substituted with the content from stdin.
Note that the tag name should not conflict with others non-replaced args.
And the -x and -X options are conflicting.
Some usages:
```
[root]# echo mypasswd | src/redis-cli -X passwd_tag mset username myname password passwd_tag OK
[root]# echo username > username.txt
[root]# head -c -1 username.txt | src/redis-cli -X name_tag mget name_tag password
1) "myname"
2) "mypasswd\n"
```
2. Handle the combination of both `-x` and `--cluster` or `-X` and `--cluster`
Extend the broadcast option to receive the last arg or <tag> arg from the stdin.
Now we can use `redis-cli -x --cluster call <host>:<port> cmd`,
or `redis-cli -X <tag> --cluster call <host>:<port> cmd <tag>`.
(support part of #9899)
issue started failing after #9878 was merged (made an exiting test more sensitive)
looks like #9982 didn't help, tested this one and it seems to work better.
this commit does two things:
1. reduce the extra delay i added earlier and instead add more keys, the effect no duration
of replication is the same, but the intervals in which the server is responsive to the tcl client is higher.
2. improve the test infra to print context when assert_error fails.
## background
Till now CONFIG SET was blocked during loading.
(In the not so distant past, GET was disallowed too)
We recently (not released yet) added an async-loading mode, see #9323,
and during that time it'll serve CONFIG SET and any other command.
And now we realized (#9770) that some configs, and commands are dangerous
during async-loading.
## changes
* Allow most CONFIG SET during loading (both on async-loading and normal loading)
* Allow CONFIG REWRITE and CONFIG RESETSTAT during loading
* Block a few config during loading (`appendonly`, `repl-diskless-load`, and `dir`)
* Block a few commands during loading (list below)
## the blocked commands:
* SAVE - obviously we don't wanna start a foregreound save during loading 8-)
* BGSAVE - we don't mind to schedule one, but we don't wanna fork now
* BGREWRITEAOF - we don't mind to schedule one, but we don't wanna fork now
* MODULE - we obviously don't wanna unload a module during replication / rdb loading
(MODULE HELP and MODULE LIST are not blocked)
* SYNC / PSYNC - we're in the middle of RDB loading from master, must not allow sync
requests now.
* REPLICAOF / SLAVEOF - we're in the middle of replicating, maybe it makes sense to let
the user abort it, but he couldn't do that so far, i don't wanna take any risk of bugs due to odd state.
* CLUSTER - only allow [HELP, SLOTS, NODES, INFO, MYID, LINKS, KEYSLOT, COUNTKEYSINSLOT,
GETKEYSINSLOT, RESET, REPLICAS, COUNT_FAILURE_REPORTS], for others, preserve the status quo
## other fixes
* processEventsWhileBlocked had an issue when being nested, this could happen with a busy script
during async loading (new), but also in a busy script during AOF loading (old). this lead to a crash in
the scenario described in #6988
# Background
The main goal of this PR is to remove relevant logics on Lua script verbatim replication,
only keeping effects replication logic, which has been set as default since Redis 5.0.
As a result, Lua in Redis 7.0 would be acting the same as Redis 6.0 with default
configuration from users' point of view.
There are lots of reasons to remove verbatim replication.
Antirez has listed some of the benefits in Issue #5292:
>1. No longer need to explain to users side effects into scripts.
They can do whatever they want.
>2. No need for a cache about scripts that we sent or not to the slaves.
>3. No need to sort the output of certain commands inside scripts
(SMEMBERS and others): this both simplifies and gains speed.
>4. No need to store scripts inside the RDB file in order to startup correctly.
>5. No problems about evicting keys during the script execution.
When looking back at Redis 5.0, antirez and core team decided to set the config
`lua-replicate-commands yes` by default instead of removing verbatim replication
directly, in case some bad situations happened. 3 years later now before Redis 7.0,
it's time to remove it formally.
# Changes
- configuration for lua-replicate-commands removed
- created config file stub for backward compatibility
- Replication script cache removed
- this is useless under script effects replication
- relevant statistics also removed
- script persistence in RDB files is also removed
- Propagation of SCRIPT LOAD and SCRIPT FLUSH to replica / AOF removed
- Deterministic execution logic in scripts removed (i.e. don't run write commands
after random ones, and sorting output of commands with random order)
- the flags indicating which commands have non-deterministic results are kept as hints to clients.
- `redis.replicate_commands()` & `redis.set_repl()` changed
- now `redis.replicate_commands()` does nothing and return an 1
- ...and then `redis.set_repl()` can be issued before `redis.replicate_commands()` now
- Relevant TCL cases adjusted
- DEBUG lua-always-replicate-commands removed
# Other changes
- Fix a recent bug comparing CLIENT_ID_AOF to original_client->flags instead of id. (introduced in #9780)
Co-authored-by: Oran Agra <oran@redislabs.com>
- add needs:debug flag for some tests
- disable "save" in external tests (speedup?)
- use debug_digest proc instead of debug command directly so it can be skipped
- use OBJECT ENCODING instead of DEBUG OBJECT to get encoding
- add a proc for OBJECT REFCOUNT so it can be skipped
- move a bunch of tests in latency_monitor tests to happen later so that latency monitor has some values in it
- add missing close_replication_stream calls
- make sure to close the temp client if DEBUG LOG fails
When rdb creates a consumer without determining whether it exists in advance,
it may return NULL and crash if it encounters corrupt data with duplicate consumers.
A test failure was reported in Daily CI.
`Crash report generated on SIGABRT` with FreeBSD.
```
*** [err]: Crash report generated on SIGABRT in tests/integration/logging.tcl
Expected [string match *crashed by signal* ### Starting...(logs) in tests/integration/logging.tcl]
```
It look like `tail -1000` was executed too early, before it
printed out all the crash logs. We can give it a few more
chances by using `wait_for_log_messages`.
Other changes:
1. In `Server is able to generate a stack trace on selected systems`,
use `wait_for_log_messages`to reduce the lines of code. And if it
fails, there are more detailed logs that can be printed.
2. In `Crash report generated on DEBUG SEGFAULT`, we also use
`wait_for_log_messages` to avoid possible timing issues.
Redis function unit is located inside functions.c
and contains Redis Function implementation:
1. FUNCTION commands:
* FUNCTION CREATE
* FCALL
* FCALL_RO
* FUNCTION DELETE
* FUNCTION KILL
* FUNCTION INFO
2. Register engine
In addition, this commit introduce the first engine
that uses the Redis Function capabilities, the
Lua engine.
Writable replicas now no longer use the values of expired keys. Expired keys are
deleted when lookupKeyWrite() is used, even on a writable replica. Previously,
writable replicas could use the value of an expired key in write commands such
as INCR, SUNIONSTORE, etc..
This commit also sorts out the mess around the functions lookupKeyRead() and
lookupKeyWrite() so they now indicate what we intend to do with the key and
are not affected by the command calling them.
Multi-key commands like SUNIONSTORE, ZUNIONSTORE, COPY and SORT with the
store option now use lookupKeyRead() for the keys they're reading from (which will
not allow reading from logically expired keys).
This commit also fixes a bug where PFCOUNT could return a value of an
expired key.
Test modules commands have their readonly and write flags updated to correctly
reflect their lookups for reading or writing. Modules are not required to
correctly reflect this in their command flags, but this change is made for
consistency since the tests serve as usage examples.
Fixes#6842. Fixes#7475.
Part three of implementing #8702, following #8887 and #9366 .
## Description of the feature
1. Replace the ziplist container of quicklist with listpack.
2. Convert existing quicklist ziplists on RDB loading time. an O(n) operation.
## Interface changes
1. New `list-max-listpack-size` config is an alias for `list-max-ziplist-size`.
2. Replace `debug ziplist` command with `debug listpack`.
## Internal changes
1. Add `lpMerge` to merge two listpacks . (same as `ziplistMerge`)
2. Add `lpRepr` to print info of listpack which is used in debugCommand and `quicklistRepr`. (same as `ziplistRepr`)
3. Replace `QUICKLIST_NODE_CONTAINER_ZIPLIST` with `QUICKLIST_NODE_CONTAINER_PACKED`(following #9357 ).
It represent that a quicklistNode is a packed node, as opposed to a plain node.
4. Remove `createZiplistObject` method, which is never used.
5. Calculate listpack entry size using overhead overestimation in `quicklistAllowInsert`.
We prefer an overestimation, which would at worse lead to a few bytes below the lowest limit of 4k.
## Improvements
1. Calling `lpShrinkToFit` after converting Ziplist to listpack, which was missed at #9366.
2. Optimize `quicklistAppendPlainNode` to avoid memcpy data.
## Bugfix
1. Fix crash in `quicklistRepr` when ziplist is compressed, introduced from #9366.
## Test
1. Add unittest for `lpMerge`.
2. Modify the old quicklist ziplist corrupt dump test.
Co-authored-by: Oran Agra <oran@redislabs.com>
In #9323, when `repl-diskless-load` is enabled and set to `swapdb`,
if the master replication ID hasn't changed, we can load data-set
asynchronously, and serving read commands during the full resync.
In `diskless loading short read` test, after a loading successfully,
we will wait for the loading to stop and continue the for loop.
After the introduction of `async_loading`, we also need to check it.
Otherwise the next loop will start too soon, may trigger a timing issue.
Issue found by corrupt-dump-fuzzer test with ASAN.
The problem was that lpSkip and lpGetWithSize could read the next listpack entry without validating that it's in range.
Similarly even the memcmp in lpFind could do that and possibly crash on segfault and now they'll crash on assert first.
The naive fix of using lpAssertValidEntry every time, resulted in 30% degradation in the lpFind benchmark of the unit test.
The final fix with the condition at the bottom has no performance implications.
TCL8.5 can't handle cases where part of the string is escaped and part of it isn't,
if there's a single char that needs escaping, we need to escape the whole string.
Leak found by the corrupt-dump-fuzzer when using GCC ASAN, which seems
to falsely report leaks on pointers kept only on the stack when calling exit.
Instead we now use _exit on panic / assert to skip these leak checks.
Additionally, check for sanitizer warnings in the corrupt-dump-fuzzer between iterations,
so that when something is found we know which test to relate it too (and it prints reproduction command list)
LCS can allocate immense amount of memory (sizes of two inputs multiplied by each other).
In the past this caused some possible security issues due to overflows, which we solved
and also added use of `trymalloc` to return "Insufficient memory" instead of OOM panic zmalloc.
But in case overcommit is enabled, it could be that we won't get the OOM panic, and zmalloc
will succeed, and then we can get OOM killed by the kernel.
The solution here is to prevent LCS from allocating transient memory that's bigger than
`proto-max-bulk-len` config.
This config is not directly related to transient memory, but using a hard coded value ad well as
introducing a specific config seems wrong.
This comes to solve an error in the corrupt-dump-fuzzer test that started in the daily CI see #9799
- Added sanitizer support. `address`, `undefined` and `thread` sanitizers are available.
- To build Redis with desired sanitizer : `make SANITIZER=undefined`
- There were some sanitizer findings, cleaned up codebase
- Added tests with address and undefined behavior sanitizers to daily CI.
- Added tests with address sanitizer to the per-PR CI (smoke out mem leaks sooner).
Basically, there are three types of issues :
**1- Unaligned load/store** : Most probably, this issue may cause a crash on a platform that
does not support unaligned access. Redis does unaligned access only on supported platforms.
**2- Signed integer overflow.** Although, signed overflow issue can be problematic time to time
and change how compiler generates code, current findings mostly about signed shift or simple
addition overflow. For most platforms Redis can be compiled for, this wouldn't cause any issue
as far as I can tell (checked generated code on godbolt.org).
**3 -Minor leak** (redis-cli), **use-after-free**(just before calling exit());
UB means nothing guaranteed and risky to reason about program behavior but I don't think any
of the fixes here worth backporting. As sanitizers are now part of the CI, preventing new issues
will be the real benefit.
First, avoid using --accurate on the freebsd CI, we only care about
systematic issues there due to being different platform, but not
accuracy
Secondly, when looking at the test which timed out it seems silly and
outdated:
- it used KEYS to attempt to trigger lazy expiry, but KEYS doesn't do
that anymore.
- it used some hard coded sleeps rather than waiting for things to
happen and exiting ASAP
In both tests, "diskless loading short read" and "diskless loading short read with module",
the timeout of waiting for the replica to respond to a short read and log it, is too short.
Also, add --dump-logs in runtest-moduleapi for valgrind runs.
For diskless replication in swapdb mode, considering we already spend replica memory
having a backup of current db to restore in case of failure, we can have the following benefits
by instead swapping database only in case we succeeded in transferring db from master:
- Avoid `LOADING` response during failed and successful synchronization for cases where the
replica is already up and running with data.
- Faster total time of diskless replication, because now we're moving from Transfer + Flush + Load
time to Transfer + Load only. Flushing the tempDb is done asynchronously after swapping.
- This could be implemented also for disk replication with similar benefits if consumers are willing
to spend the extra memory usage.
General notes:
- The concept of `backupDb` becomes `tempDb` for clarity.
- Async loading mode will only kick in if the replica is syncing from a master that has the same
repl-id the one it had before. i.e. the data it's getting belongs to a different time of the same timeline.
- New property in INFO: `async_loading` to differentiate from the blocking loading
- Slot to Key mapping is now a field of `redisDb` as it's more natural to access it from both server.db
and the tempDb that is passed around.
- Because this is affecting replicas only, we assume that if they are not readonly and write commands
during replication, they are lost after SYNC same way as before, but we're still denying CONFIG SET
here anyways to avoid complications.
Considerations for review:
- We have many cases where server.loading flag is used and even though I tried my best, there may
be cases where async_loading should be checked as well and cases where it shouldn't (would require
very good understanding of whole code)
- Several places that had different behavior depending on the loading flag where actually meant to just
handle commands coming from the AOF client differently than ones coming from real clients, changed
to check CLIENT_ID_AOF instead.
**Additional for Release Notes**
- Bugfix - server.dirty was not incremented for any kind of diskless replication, as effect it wouldn't
contribute on triggering next database SAVE
- New flag for RM_GetContextFlags module API: REDISMODULE_CTX_FLAGS_ASYNC_LOADING
- Deprecated RedisModuleEvent_ReplBackup. Starting from Redis 7.0, we don't fire this event.
Instead, we have the new RedisModuleEvent_ReplAsyncLoad holding 3 sub-events: STARTED,
ABORTED and COMPLETED.
- New module flag REDISMODULE_OPTIONS_HANDLE_REPL_ASYNC_LOAD for RedisModule_SetModuleOptions
to allow modules to declare they support the diskless replication with async loading (when absent, we fall
back to disk-based loading).
Co-authored-by: Eduardo Semprebon <edus@saxobank.com>
Co-authored-by: Oran Agra <oran@redislabs.com>
When repl-diskless-load is enabled, the connection is set to the blocking state.
The connection may be interrupted by a signal during a system call.
This would have resulted in a disconnection and possibly a reconnection loop.
Co-authored-by: Oran Agra <oran@redislabs.com>
Redis lists are stored in quicklist, which is currently a linked list of ziplists.
Ziplists are limited to storing elements no larger than 4GB, so when bigger
items are added they're getting truncated.
This PR changes quicklists so that they're capable of storing large items
in quicklist nodes that are plain string buffers rather than ziplist.
As part of the PR there were few other changes in redis:
1. new DEBUG sub-commands:
- QUICKLIST-PACKED-THRESHOLD - set the threshold of for the node type to
be plan or ziplist. default (1GB)
- QUICKLIST <key> - Shows low level info about the quicklist encoding of <key>
2. rdb format change:
- A new type was added - RDB_TYPE_LIST_QUICKLIST_2 .
- container type (packed / plain) was added to the beginning of the rdb object
(before the actual node list).
3. testing:
- Tests that requires over 100MB will be by default skipped. a new flag was
added to 'runtest' to run the large memory tests (not used by default)
Co-authored-by: sundb <sundbcn@gmail.com>
Co-authored-by: Oran Agra <oran@redislabs.com>
So it looks like sampling set loglines [count_log_lines -2] was
executed too late, and the replication managed to complete before that.
```
*** [err]: diskless no replicas drop during rdb pipe in tests/integration/replication.tcl
log message of '"*Diskless rdb transfer, done reading from pipe, 2 replicas still up*"' not found in ./tests/tmp/server.6124.69/stdout after line: 52 till line: 52
```
Changes:
1. when we search the master log file, we start to search from before we sent the REPLICAOF
command, to prevent a race in which the replication completed before we sampled the log line count.
2. we don't need to sample the replica loglines sine it's a fresh resplica that's just been started, so the message
we're looking for is the first occurrence in the log, we can start search from 0.
Co-authored-by: Oran Agra <oran@redislabs.com>
Test failed on freebsd:
```
*** [err]: Make the old master a replica of the new one and check conditions in tests/integration/psync2-pingoff.tcl
Expected '162' to be equal to '176' (context: type eval line 18 cmd {assert_equal [status $R(0) master_repl_offset] [status $R(1) master_repl_offset]} proc ::test)
```
There are two possible race conditions in the test.
1. The code waits for sync_full to increment, and assumes that means the
master did the fork. But in fact there are cases the master will increment
that sync_full counter (after replica asks for sync), but will see that
there's already a fork running and will delay the fork creation.
In this case the INCR will be executed before the fork happens, so it'll
not be in the command stream. Solve that by waiting for `master_link_status: up`
on the replica before the INCR.
2. The repl-ping-replica-period is still high (1 second), so there's a chance the
master will send an additional PING between the two calls to INFO (the line that
fails is the one that samples INFO from both servers). So there's a chance one of
them will have an incremented offset due to PING and the other won't have it yet.
In theory we can wait for the repl_offset to match, but then we risk facing a
situation where that race will hide an offset mis-match. so instead, i think we
should just change repl-ping-replica-period to prevent further pings from being pushed.
Co-authored-by: Oran Agra <oran@redislabs.com>
Add timestamp annotation in AOF, one part of #9325.
Enabled with the new `aof-timestamp-enabled` config option.
Timestamp annotation format is "#TS:${timestamp}\r\n"."
TS" is short of timestamp and this method could save extra bytes in AOF.
We can use timestamp annotation for some special functions.
- know the executing time of commands
- restore data to a specific point-in-time (by using redis-check-rdb to truncate the file)
## Background
For redis master, one replica uses one copy of replication buffer, that is a big waste of memory,
more replicas more waste, and allocate/free memory for every reply list also cost much.
If we set client-output-buffer-limit small and write traffic is heavy, master may disconnect with
replicas and can't finish synchronization with replica. If we set client-output-buffer-limit big,
master may be OOM when there are many replicas that separately keep much memory.
Because replication buffers of different replica client are the same, one simple idea is that
all replicas only use one replication buffer, that will effectively save memory.
Since replication backlog content is the same as replicas' output buffer, now we
can discard replication backlog memory and use global shared replication buffer
to implement replication backlog mechanism.
## Implementation
I create one global "replication buffer" which contains content of replication stream.
The structure of "replication buffer" is similar to the reply list that exists in every client.
But the node of list is `replBufBlock`, which has `id, repl_offset, refcount` fields.
```c
/* Replication buffer blocks is the list of replBufBlock.
*
* +--------------+ +--------------+ +--------------+
* | refcount = 1 | ... | refcount = 0 | ... | refcount = 2 |
* +--------------+ +--------------+ +--------------+
* | / \
* | / \
* | / \
* Repl Backlog Replia_A Replia_B
*
* Each replica or replication backlog increments only the refcount of the
* 'ref_repl_buf_node' which it points to. So when replica walks to the next
* node, it should first increase the next node's refcount, and when we trim
* the replication buffer nodes, we remove node always from the head node which
* refcount is 0. If the refcount of the head node is not 0, we must stop
* trimming and never iterate the next node. */
/* Similar with 'clientReplyBlock', it is used for shared buffers between
* all replica clients and replication backlog. */
typedef struct replBufBlock {
int refcount; /* Number of replicas or repl backlog using. */
long long id; /* The unique incremental number. */
long long repl_offset; /* Start replication offset of the block. */
size_t size, used;
char buf[];
} replBufBlock;
```
So now when we feed replication stream into replication backlog and all replicas, we only need
to feed stream into replication buffer `feedReplicationBuffer`. In this function, we set some fields of
replication backlog and replicas to references of the global replication buffer blocks. And we also
need to check replicas' output buffer limit to free if exceeding `client-output-buffer-limit`, and trim
replication backlog if exceeding `repl-backlog-size`.
When sending reply to replicas, we also need to iterate replication buffer blocks and send its
content, when totally sending one block for replica, we decrease current node count and
increase the next current node count, and then free the block which reference is 0 from the
head of replication buffer blocks.
Since now we use linked list to manage replication backlog, it may cost much time for iterating
all linked list nodes to find corresponding replication buffer node. So we create a rax tree to
store some nodes for index, but to avoid rax tree occupying too much memory, i record
one per 64 nodes for index.
Currently, to make partial resynchronization as possible as much, we always let replication
backlog as the last reference of replication buffer blocks, backlog size may exceeds our setting
if slow replicas that reference vast replication buffer blocks, and this method doesn't increase
memory usage since they share replication buffer. To avoid freezing server for freeing unreferenced
replication buffer blocks when we need to trim backlog for exceeding backlog size setting,
we trim backlog incrementally (free 64 blocks per call now), and make it faster in
`beforeSleep` (free 640 blocks).
### Other changes
- `mem_total_replication_buffers`: we add this field in INFO command, it means the total
memory of replication buffers used.
- `mem_clients_slaves`: now even replica is slow to replicate, and its output buffer memory
is not 0, but it still may be 0, since replication backlog and replicas share one global replication
buffer, only if replication buffer memory is more than the repl backlog setting size, we consider
the excess as replicas' memory. Otherwise, we think replication buffer memory is the consumption
of repl backlog.
- Key eviction
Since all replicas and replication backlog share global replication buffer, we think only the
part of exceeding backlog size the extra separate consumption of replicas.
Because we trim backlog incrementally in the background, backlog size may exceeds our
setting if slow replicas that reference vast replication buffer blocks disconnect.
To avoid massive eviction loop, we don't count the delayed freed replication backlog into
used memory even if there are no replicas, i.e. we also regard this memory as replicas's memory.
- `client-output-buffer-limit` check for replica clients
It doesn't make sense to set the replica clients output buffer limit lower than the repl-backlog-size
config (partial sync will succeed and then replica will get disconnected). Such a configuration is
ignored (the size of repl-backlog-size will be used). This doesn't have memory consumption
implications since the replica client will share the backlog buffers memory.
- Drop replication backlog after loading data if needed
We always create replication backlog if server is a master, we need it because we put DELs in
it when loading expired keys in RDB, but if RDB doesn't have replication info or there is no rdb,
it is not possible to support partial resynchronization, to avoid extra memory of replication backlog,
we drop it.
- Multi IO threads
Since all replicas and replication backlog use global replication buffer, if I/O threads are enabled,
to guarantee data accessing thread safe, we must let main thread handle sending the output buffer
to all replicas. But before, other IO threads could handle sending output buffer of all replicas.
## Other optimizations
This solution resolve some other problem:
- When replicas disconnect with master since of out of output buffer limit, releasing the output
buffer of replicas may freeze server if we set big `client-output-buffer-limit` for replicas, but now,
it doesn't cause freezing.
- This implementation may mitigate reply list copy cost time(also freezes server) when one replication
has huge reply buffer and another replica can copy buffer for full synchronization. now, we just copy
reference info, it is very light.
- If we set replication backlog size big, it also may cost much time to copy replication backlog into
replica's output buffer. But this commit eliminates this problem.
- Resizing replication backlog size doesn't empty current replication backlog content.
in the past few days i've seen two failures in the valgrind daily test.
*** [err]: slave fails full sync and diskless load swapdb recovers it in tests/integration/replication.tcl
Replica didn't get into loading mode
can't reproduce it, but i'm hoping it's just too slow (to start loading within 5 seconds)
Since we measure the COW size in this test by changing some keys and reading
the reported COW size, we need to ensure that the "dismiss mechanism" (#8974)
will not free memory and reduce the COW size.
For that, this commit changes the size of the keys to 512B (less than a page).
and because some keys may fall into the same page, we are modifying ten keys
on each iteration and check for at least 50% change in the COW size.
This is similar to the recent addition of LMPOP/BLMPOP (#9373), but zset.
Syntax for the new ZMPOP command:
`ZMPOP numkeys [<key> ...] MIN|MAX [COUNT count]`
Syntax for the new BZMPOP command:
`BZMPOP timeout numkeys [<key> ...] MIN|MAX [COUNT count]`
Some background:
- ZPOPMIN/ZPOPMAX take only one key, and can return multiple elements.
- BZPOPMIN/BZPOPMAX take multiple keys, but return only one element from just one key.
- ZMPOP/BZMPOP can take multiple keys, and can return multiple elements from just one key.
Note that ZMPOP/BZMPOP can take multiple keys, it eventually operates on just on key.
And it will propagate as ZPOPMIN or ZPOPMAX with the COUNT option.
As new commands, if we can not pop any elements, the response like:
- ZMPOP: Return a NIL in both RESP2 and RESP3, unlike ZPOPMIN/ZPOPMAX return emptyarray.
- BZMPOP: Return a NIL in both RESP2 and RESP3 when timeout is reached, like BZPOPMIN/BZPOPMAX.
For the normal response is nested arrays in RESP2 and RESP3:
```
ZMPOP/BZMPOP
1) keyname
2) 1) 1) member1
2) score1
2) 1) member2
2) score2
In RESP2:
1) "myzset"
2) 1) 1) "three"
2) "3"
2) 1) "two"
2) "2"
In RESP3:
1) "myzset"
2) 1) 1) "three"
2) (double) 3
2) 1) "two"
2) (double) 2
```
- Add `-u <uri>` command line option to support `redis://` URI scheme.
- included server connection information object (`struct cliConnInfo`),
used to describe an ip:port pair, db num user input, and user:pass to
avoid a large number of function arguments.
- Using sds on connection info strings for redis-benchmark/redis-cli
Co-authored-by: yoav-steinberg <yoav@monfort.co.il>
The main idea is how to allow a master to load replication info from RDB file when rebooting, if master can load replication info it means that replicas may have the chance to psync with master, it can save much traffic.
The key point is we need guarantee safety and consistency, so there
are two differences between master and replica:
1. master would load the replication info as secondary ID and
offset, in case other masters have the same replid.
2. when master loading RDB, it would propagate expired keys as DEL
command to replication backlog, then replica can receive these
commands to delete stale keys.
p.s. the expired keys when RDB loading is useful for users, so
we show it as `rdb_last_load_keys_expired` and `rdb_last_load_keys_loaded` in info persistence.
Moreover, after load replication info, master should update
`no_replica_time` in case loading RDB cost too long time.
Part two of implementing #8702 (zset), after #8887.
## Description of the feature
Replaced all uses of ziplist with listpack in t_zset, and optimized some of the code to optimize performance.
## Rdb format changes
New `RDB_TYPE_ZSET_LISTPACK` rdb type.
## Rdb loading improvements:
1) Pre-expansion of dict for validation of duplicate data for listpack and ziplist.
2) Simplifying the release of empty key objects when RDB loading.
3) Unify ziplist and listpack data verify methods for zset and hash, and move code to rdb.c.
## Interface changes
1) New `zset-max-listpack-entries` config is an alias for `zset-max-ziplist-entries` (same with `zset-max-listpack-value`).
2) OBJECT ENCODING will return listpack instead of ziplist.
## Listpack improvements:
1) Add `lpDeleteRange` and `lpDeleteRangeWithEntry` functions to delete a range of entries from listpack.
2) Improve the performance of `lpCompare`, converting from string to integer is faster than converting from integer to string.
3) Replace `snprintf` with `ll2string` to improve performance in converting numbers to strings in `lpGet()`.
## Zset improvements:
1) Improve the performance of `zzlFind` method, use `lpFind` instead of `lpCompare` in a loop.
2) Use `lpDeleteRangeWithEntry` instead of `lpDelete` twice to delete a element of zset.
## Tests
1) Add some unittests for `lpDeleteRange` and `lpDeleteRangeWithEntry` function.
2) Add zset RDB loading test.
3) Add benchmark test for `lpCompare` and `ziplsitCompare`.
4) Add empty listpack zset corrupt dump test.
We want to add COUNT option for BLPOP.
But we can't do it without breaking compatibility due to the command arguments syntax.
So this commit introduce two new commands.
Syntax for the new LMPOP command:
`LMPOP numkeys [<key> ...] LEFT|RIGHT [COUNT count]`
Syntax for the new BLMPOP command:
`BLMPOP timeout numkeys [<key> ...] LEFT|RIGHT [COUNT count]`
Some background:
- LPOP takes one key, and can return multiple elements.
- BLPOP takes multiple keys, but returns one element from just one key.
- LMPOP can take multiple keys and return multiple elements from just one key.
Note that LMPOP/BLMPOP can take multiple keys, it eventually operates on just one key.
And it will propagate as LPOP or RPOP with the COUNT option.
As a new command, it still return NIL if we can't pop any elements.
For the normal response is nested arrays in RESP2 and RESP3, like:
```
LMPOP/BLMPOP
1) keyname
2) 1) element1
2) element2
```
I.e. unlike BLPOP that returns a key name and one element so it uses a flat array,
and LPOP that returns multiple elements with no key name, and again uses a flat array,
this one has to return a nested array, and it does for for both RESP2 and RESP3 (like SCAN does)
Some discuss can see: #766#8824
* Delay to discard cache master when full synchronization
* Don't disconnect with replicas before loading transferred RDB when full sync
Previously, once replica need to start full synchronization with master,
it will discard cached master whatever full synchronization is failed or
not.
Now we discard cached master only when transferring RDB is finished
and start to change data space, this make replica could start partial
resynchronization with another new master if new master is failed
during full synchronization.
Until now, giving a negative index seeks from the end of a list and a
positive seeks from the beginning. This change makes it seek from
the nearest end, regardless of the sign of the given index.
quicklistIndex is used by all list commands which operate by index.
LINDEX key 999999 in a list if 1M elements is greately optimized by
this change. Latency is cut by 75%.
LINDEX key -1000000 in a list of 1M elements, likewise.
LRANGE key -1 -1 is affected by this, since LRANGE converts the
indices to positive numbers before seeking.
The tests for corrupt dumps are updated to make sure the corrup
data is seeked in the same direction as before.
1. The output of --help:
* On the Usage line, just write [OPTIONS] [COMMAND ARGS...] instead listing
only a few arbitrary options and no command.
* For --cluster, describe that if the command is supplied on the command line,
the key must contain "{tag}". Otherwise, the command will not be sent to the
right cluster node.
* For -r, add a note that if -r is omitted, all commands in a benchmark will
use the same key. Also align the description.
* For -t, describe that -t is ignored if a command is supplied on the command
line.
2. Print a warning if -t is present when a specific command is supplied.
3. Print all warnings and errors to stderr.
4. Remove -e from calls in redis-benchmark test suite.
We only run OOM related tests on x86_64 and aarch64, as jemalloc on other
platforms (notably s390x) may actually succeed very large allocations. As
a result the test may hang for a very long time at the cleanup phase,
iterating as many as 2^61 hash table slots.
