## Introduction
Redis introduced IO Thread in 6.0, allowing IO threads to handle client
request reading, command parsing and reply writing, thereby improving
performance. The current IO thread implementation has a few drawbacks.
- The main thread is blocked during IO thread read/write operations and
must wait for all IO threads to complete their current tasks before it
can continue execution. In other words, the entire process is
synchronous. This prevents the efficient utilization of multi-core CPUs
for parallel processing.
- When the number of clients and requests increases moderately, it
causes all IO threads to reach full CPU utilization due to the busy wait
mechanism used by the IO threads. This makes it challenging for us to
determine which part of Redis has reached its bottleneck.
- When IO threads are enabled with TLS and io-threads-do-reads, a
disconnection of a connection with pending data may result in it being
assigned to multiple IO threads simultaneously. This can cause race
conditions and trigger assertion failures. Related issue:
https://github.com/redis/redis/issues/12540
Therefore, we designed an asynchronous IO threads solution. The IO
threads adopt an event-driven model, with the main thread dedicated to
command processing, meanwhile, the IO threads handle client read and
write operations in parallel.
## Implementation
### Overall
As before, we did not change the fact that all client commands must be
executed on the main thread, because Redis was originally designed to be
single-threaded, and processing commands in a multi-threaded manner
would inevitably introduce numerous race and synchronization issues. But
now each IO thread has independent event loop, therefore, IO threads can
use a multiplexing approach to handle client read and write operations,
eliminating the CPU overhead caused by busy-waiting.
the execution process can be briefly described as follows:
the main thread assigns clients to IO threads after accepting
connections, IO threads will notify the main thread when clients
finish reading and parsing queries, then the main thread processes
queries from IO threads and generates replies, IO threads handle
writing reply to clients after receiving clients list from main thread,
and then continue to handle client read and write events.
### Each IO thread has independent event loop
We now assign each IO thread its own event loop. This approach
eliminates the need for the main thread to perform the costly
`epoll_wait` operation for handling connections (except for specific
ones). Instead, the main thread processes requests from the IO threads
and hands them back once completed, fully offloading read and write
events to the IO threads.
Additionally, all TLS operations, including handling pending data, have
been moved entirely to the IO threads. This resolves the issue where
io-threads-do-reads could not be used with TLS.
### Event-notified client queue
To facilitate communication between the IO threads and the main thread,
we designed an event-notified client queue. Each IO thread and the main
thread have two such queues to store clients waiting to be processed.
These queues are also integrated with the event loop to enable handling.
We use pthread_mutex to ensure the safety of queue operations, as well
as data visibility and ordering, and race conditions are minimized, as
each IO thread and the main thread operate on independent queues,
avoiding thread suspension due to lock contention. And we implemented an
event notifier based on `eventfd` or `pipe` to support event-driven
handling.
### Thread safety
Since the main thread and IO threads can execute in parallel, we must
handle data race issues carefully.
**client->flags**
The primary tasks of IO threads are reading and writing, i.e.
`readQueryFromClient` and `writeToClient`. However, IO threads and the
main thread may concurrently modify or access `client->flags`, leading
to potential race conditions. To address this, we introduced an io-flags
variable to record operations performed by IO threads, thereby avoiding
race conditions on `client->flags`.
**Pause IO thread**
In the main thread, we may want to operate data of IO threads, maybe
uninstall event handler, access or operate query/output buffer or resize
event loop, we need a clean and safe context to do that. We pause IO
thread in `IOThreadBeforeSleep`, do some jobs and then resume it. To
avoid thread suspended, we use busy waiting to confirm the target
status. Besides we use atomic variable to make sure memory visibility
and ordering. We introduce these functions to pause/resume IO Threads as
below.
```
pauseIOThread, resumeIOThread
pauseAllIOThreads, resumeAllIOThreads
pauseIOThreadsRange, resumeIOThreadsRange
```
Testing has shown that `pauseIOThread` is highly efficient, allowing the
main thread to execute nearly 200,000 operations per second during
stress tests. Similarly, `pauseAllIOThreads` with 8 IO threads can
handle up to nearly 56,000 operations per second. But operations
performed between pausing and resuming IO threads must be quick;
otherwise, they could cause the IO threads to reach full CPU
utilization.
**freeClient and freeClientAsync**
The main thread may need to terminate a client currently running on an
IO thread, for example, due to ACL rule changes, reaching the output
buffer limit, or evicting a client. In such cases, we need to pause the
IO thread to safely operate on the client.
**maxclients and maxmemory-clients updating**
When adjusting `maxclients`, we need to resize the event loop for all IO
threads. Similarly, when modifying `maxmemory-clients`, we need to
traverse all clients to calculate their memory usage. To ensure safe
operations, we pause all IO threads during these adjustments.
**Client info reading**
The main thread may need to read a client’s fields to generate a
descriptive string, such as for the `CLIENT LIST` command or logging
purposes. In such cases, we need to pause the IO thread handling that
client. If information for all clients needs to be displayed, all IO
threads must be paused.
**Tracking redirect**
Redis supports the tracking feature and can even send invalidation
messages to a connection with a specified ID. But the target client may
be running on IO thread, directly manipulating the client’s output
buffer is not thread-safe, and the IO thread may not be aware that the
client requires a response. In such cases, we pause the IO thread
handling the client, modify the output buffer, and install a write event
handler to ensure proper handling.
**clientsCron**
In the `clientsCron` function, the main thread needs to traverse all
clients to perform operations such as timeout checks, verifying whether
they have reached the soft output buffer limit, resizing the
output/query buffer, or updating memory usage. To safely operate on a
client, the IO thread handling that client must be paused.
If we were to pause the IO thread for each client individually, the
efficiency would be very low. Conversely, pausing all IO threads
simultaneously would be costly, especially when there are many IO
threads, as clientsCron is invoked relatively frequently.
To address this, we adopted a batched approach for pausing IO threads.
At most, 8 IO threads are paused at a time. The operations mentioned
above are only performed on clients running in the paused IO threads,
significantly reducing overhead while maintaining safety.
### Observability
In the current design, the main thread always assigns clients to the IO
thread with the least clients. To clearly observe the number of clients
handled by each IO thread, we added the new section in INFO output. The
`INFO THREADS` section can show the client count for each IO thread.
```
# Threads
io_thread_0:clients=0
io_thread_1:clients=2
io_thread_2:clients=2
```
Additionally, in the `CLIENT LIST` output, we also added a field to
indicate the thread to which each client is assigned.
`id=244 addr=127.0.0.1:41870 laddr=127.0.0.1:6379 ... resp=2 lib-name=
lib-ver= io-thread=1`
## Trade-off
### Special Clients
For certain special types of clients, keeping them running on IO threads
would result in severe race issues that are difficult to resolve.
Therefore, we chose not to offload these clients to the IO threads.
For replica, monitor, subscribe, and tracking clients, main thread may
directly write them a reply when conditions are met. Race issues are
difficult to resolve, so we have them processed in the main thread. This
includes the Lua debug clients as well, since we may operate connection
directly.
For blocking client, after the IO thread reads and parses a command and
hands it over to the main thread, if the client is identified as a
blocking type, it will be remained in the main thread. Once the blocking
operation completes and the reply is generated, the client is
transferred back to the IO thread to send the reply and wait for event
triggers.
### Clients Eviction
To support client eviction, it is necessary to update each client’s
memory usage promptly during operations such as read, write, or command
execution. However, when a client operates on an IO thread, it is not
feasible to update the memory usage immediately due to the risk of data
races. As a result, memory usage can only be updated either in the main
thread while processing commands or in the `ClientsCron` periodically.
The downside of this approach is that updates might experience a delay
of up to one second, which could impact the precision of memory
management for eviction.
To avoid incorrectly evicting clients. We adopted a best-effort
compensation solution, when we decide to eviction a client, we update
its memory usage again before evicting, if the memory used by the client
does not decrease or memory usage bucket is not changed, then we will
evict it, otherwise, not evict it.
However, we have not completely solved this problem. Due to the delay in
memory usage updates, it may lead us to make incorrect decisions about
the need to evict clients.
### Defragment
In the majority of cases we do NOT use the data from argv directly in
the db.
1. key names
We store a copy that we allocate in the main thread, see `sdsdup()` in
`dbAdd()`.
2. hash key and value
We store key as hfield and store value as sds, see `hfieldNew()` and
`sdsdup()` in `hashTypeSet()`.
3. other datatypes
They don't even use SDS, so there is no reference issues.
But in some cases client the data from argv may be retain by the main
thread.
As a result, during fragmentation cleanup, we need to move allocations
from the IO thread’s arena to the main thread’s arena. We always
allocate new memory in the main thread’s arena, but the memory released
by IO threads may not yet have been reclaimed. This ultimately causes
the fragmentation rate to be higher compared to creating and allocating
entirely within a single thread.
The following cases below will lead to memory allocated by the IO thread
being kept by the main thread.
1. string related command: `append`, `getset`, `mset` and `set`.
If `tryObjectEncoding()` does not change argv, we will keep it directly
in the main thread, see the code in `tryObjectEncoding()`(specifically
`trimStringObjectIfNeeded()`)
2. block related command.
the key names will be kept in `c->db->blocking_keys`.
3. watch command
the key names will be kept in `c->db->watched_keys`.
4. [s]subscribe command
channel name will be kept in `serverPubSubChannels`.
5. script load command
script will be kept in `server.lua_scripts`.
7. some module API: `RM_RetainString`, `RM_HoldString`
Those issues will be handled in other PRs.
## Testing
### Functional Testing
The commit with enabling IO Threads has passed all TCL tests, but we did
some changes:
**Client query buffer**: In the original code, when using a reusable
query buffer, ownership of the query buffer would be released after the
command was processed. However, with IO threads enabled, the client
transitions from an IO thread to the main thread for processing. This
causes the ownership release to occur earlier than the command
execution. As a result, when IO threads are enabled, the client's
information will never indicate that a shared query buffer is in use.
Therefore, we skip the corresponding query buffer tests in this case.
**Defragment**: Add a new defragmentation test to verify the effect of
io threads on defragmentation.
**Command delay**: For deferred clients in TCL tests, due to clients
being assigned to different threads for execution, delays may occur. To
address this, we introduced conditional waiting: the process proceeds to
the next step only when the `client list` contains the corresponding
commands.
### Sanitizer Testing
The commit passed all TCL tests and reported no errors when compiled
with the `fsanitizer=thread` and `fsanitizer=address` options enabled.
But we made the following modifications: we suppressed the sanitizer
warnings for clients with watched keys when updating `client->flags`, we
think IO threads read `client->flags`, but never modify it or read the
`CLIENT_DIRTY_CAS` bit, main thread just only modifies this bit, so
there is no actual data race.
## Others
### IO thread number
In the new multi-threaded design, the main thread is primarily focused
on command processing to improve performance. Typically, the main thread
does not handle regular client I/O operations but is responsible for
clients such as replication and tracking clients. To avoid breaking
changes, we still consider the main thread as the first IO thread.
When the io-threads configuration is set to a low value (e.g., 2),
performance does not show a significant improvement compared to a
single-threaded setup for simple commands (such as SET or GET), as the
main thread does not consume much CPU for these simple operations. This
results in underutilized multi-core capacity. However, for more complex
commands, having a low number of IO threads may still be beneficial.
Therefore, it’s important to adjust the `io-threads` based on your own
performance tests.
Additionally, you can clearly monitor the CPU utilization of the main
thread and IO threads using `top -H -p $redis_pid`. This allows you to
easily identify where the bottleneck is. If the IO thread is the
bottleneck, increasing the `io-threads` will improve performance. If the
main thread is the bottleneck, the overall performance can only be
scaled by increasing the number of shards or replicas.
---------
Co-authored-by: debing.sun <debing.sun@redis.com>
Co-authored-by: oranagra <oran@redislabs.com>
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>
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].
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.
There are situations (especially in TLS) in which the engine gets too occupied managing a large number of new connections. Existing connections may time-out while the server is processing the new connections initial TLS handshakes, which may cause cause new connections to be established, perpetuating the problem. To better manage the tradeoff between new connection rate and other workloads, this change adds a new config to manage maximum number of new connections per event loop cycle, instead of using a predetermined number (currently 1000).
This change introduces two new configurations, max-new-connections-per-cycle and max-new-tls-connections-per-cycle. The default value of the tcp connections is 10 per cycle and the default value of tls connections per cycle is 1.
---------
Co-authored-by: Madelyn Olson <madelyneolson@gmail.com>
This is a follow-up fix to #12733. We need to apply the same changes to
delKeysInSlot. Refer to #12733 for more details.
This PR contains some other minor cleanups / improvements to the test
suite and docs.
It uses the postnotifications test module in a cluster mode test which
revealed a leak in the test module (fixed).
The following four configurations are renamed to align with Redis style:
1. server_cpulist renamed to server-cpulist
2. bio_cpulist renamed to bio-cpulist
3. aof_rewrite_cpulist renamed to aof-rewrite-cpulist
4. bgsave_cpulist renamed to bgsave-cpulist
The original names are retained as aliases to ensure compatibility with
old configuration files. We recommend users to gradually transition to
using the new configuration names to maintain consistency in style.
warning against editing the config file and restarting the server.
which will attempt to load an AOF file and disregard the RDB.
Co-authored-by: Oran Agra <oran@redislabs.com>
This PR adds a human readable name to a node in clusters that are visible as part of error logs. This is useful so that admins and operators of Redis cluster have better visibility into failures without having to cross-reference the generated ID with some logical identifier (such as pod-ID or EC2 instance ID). This is mentioned in #8948. Specific nodenames can be set by using the variable cluster-announce-human-nodename. The nodename is gossiped using the clusterbus extension in #9530.
Co-authored-by: Madelyn Olson <madelyneolson@gmail.com>
Users can record logs of different levels by setting the `loglevel`.
However, sometimes there are many logs even at the warning level,
which can affect the performance of Redis.
For example, when a user accesses the tls-port using a non-encrypted link,
Redis will log lots of "# Error accepting a client connection: ...".
We can provide the ability to disable logging so that users can temporarily turn
off logging and turn it back on after the problem is resolved.
Small sets with not only integer elements are listpack encoded, by default
up to 128 elements, max 64 bytes per element, new config `set-max-listpack-entries`
and `set-max-listpack-value`. This saves memory for small sets compared to using a hashtable.
Sets with only integers, even very small sets, are still intset encoded (up to 1G
limit, etc.). Larger sets are hashtable encoded.
This PR increments the RDB version, and has an effect on OBJECT ENCODING
Possible conversions when elements are added:
intset -> listpack
listpack -> hashtable
intset -> hashtable
Note: No conversion happens when elements are deleted. If all elements are
deleted and then added again, the set is deleted and recreated, thus implicitly
converted to a smaller encoding.
Starting from 6.2, after ACL SETUSER user reset, the user
will carry the sanitize-payload flag. It was added in #7807,
and then ACL SETUSER reset is inconsistent with default
newly created user which missing sanitize-payload flag.
Same as `off` and `on` these two bits are mutually exclusive,
the default created user needs to have sanitize-payload flag.
Adds USER_FLAG_SANITIZE_PAYLOAD flag to ACLCreateUser.
Note that the bug don't have any real implications,
since the code in rdb.c (rdbLoadObject) checks for
`USER_FLAG_SANITIZE_PAYLOAD_SKIP`, so the fact that
`USER_FLAG_SANITIZE_PAYLOAD` is missing doesn't really matters.
Added tests to make sure it won't be broken in the future,
and updated the comment in ACLSetUser and redis.conf
Just noticed that there are some inaccurate, or at least confusing information about `repl-diskless-load` in `redis.conf`
It shouldn't scare away users willing to spend the extra memory.
`may mean that we have to flush the contents of the current database before the full rdb was received.`: this is likely related to the time when there was an option `always`, where content on replica was flushed before loading from master.
Till now Redis officially supported tuning it via environment variable see #1074.
But we had other requests to allow changing it at runtime, see #799, and #11041.
Note that `strcoll()` is used as Lua comparison function and also for comparison of
certain string objects in Redis, which leads to a problem that, in different regions,
for some characters, the result may be different. Below is an example.
```
127.0.0.1:6333> SORT test alpha
1) "<"
2) ">"
3) ","
4) "*"
127.0.0.1:6333> CONFIG GET locale-collate
1) "locale-collate"
2) ""
127.0.0.1:6333> CONFIG SET locale-collate 1
(error) ERR CONFIG SET failed (possibly related to argument 'locale')
127.0.0.1:6333> CONFIG SET locale-collate C
OK
127.0.0.1:6333> SORT test alpha
1) "*"
2) ","
3) "<"
4) ">"
```
That will cause accidental code compatibility issues for Lua scripts and some
Redis commands. This commit creates a new config parameter to control the
local environment which only affects `Collate` category. Above shows how it
affects `SORT` command, and below shows the influence on Lua scripts.
```
127.0.0.1:6333> CONFIG GET locale-collate
1) " locale-collate"
2) "C"
127.0.0.1:6333> EVAL "return ',' < '*'" 0
(nil)
127.0.0.1:6333> CONFIG SET locale-collate ""
OK
127.0.0.1:6333> EVAL "return ',' < '*'" 0
(integer) 1
```
Co-authored-by: calvincjli <calvincjli@tencent.com>
Co-authored-by: Oran Agra <oran@redislabs.com>
The divided by two and less <= 10 logics were changed in 06ca9d6839.
Now we just decrement the counter by num_periods.
The lfu-decay-time special value of 0 's meaning was actually changed in 06ca9d6839.
Now we won't do anything on counter if lfu-decay-time is 0.
The SHUTDOWN command has various flags to change it's default behavior,
but in some cases establishing a connection to redis is complicated and it's easier
for the management software to use signals. however, so far the signals could only
trigger the default shutdown behavior.
Here we introduce the option to control shutdown arguments for SIGTERM and SIGINT.
New config options:
`shutdown-on-sigint [nosave | save] [now] [force]`
`shutdown-on-sigterm [nosave | save] [now] [force]`
Implementation:
Support MULTI_ARG_CONFIG on createEnumConfig to support multiple enums to be applied as bit flags.
Co-authored-by: Oran Agra <oran@redislabs.com>
* Till now, replicas that were unable to persist, would still execute the commands
they got from the master, now they'll panic by default, and we add a new
`replica-ignore-disk-errors` config to change that.
* Till now, when a command failed on a replica or AOF-loading, it only logged a
warning and a stat, we add a new `propagation-error-behavior` config to allow
panicking in that state (may become the default one day)
Note that commands that fail on the replica can either indicate a bug that could
cause data inconsistency between the replica and the master, or they could be
in some cases (specifically in previous versions), a result of a command (e.g. EVAL)
that failed on the master, but still had to be propagated to fail on the replica as well.
Add a configuration option to attach an operating system-specific identifier to Redis sockets, supporting advanced network configurations using iptables (Linux) or ipfw (FreeBSD).
Add an optional keyspace event when new keys are added to the db.
This is useful for applications where clients need to be aware of the redis keyspace.
Such an application can SCAN once at startup and then listen for "new" events (plus
others associated with DEL, RENAME, etc).
Did some cleanups:
1. local local typo
2. replace the only slave word in the file
3. add FUNCTION FLUSH to `lazyfree-lazy-user-flush` description
4. thought it would be better to use these, there are actually "four" options
5. the the typo
6. remove a extra space
7. change comment next to `activedefrag no` to match the default value
publishshard was added in #8621 (7.0 RC1), but the publishshard_sent
stat is not shown in CLUSTER INFO command.
Other changes:
1. Remove useless `needhelp` statements, it was removed in 3dad819.
2. Use `LL_WARNING` log level for some error logs (I/O error, Connection failed).
3. Fix typos that saw by the way.
* Drop obsolete initialization calls.
* Use decoder API for DH parameters.
* Enable auto DH parameters if not explicitly used, which should be the
preferred configuration going forward.
For backwards compatibility in 6.x, channels default permission was set to `allchannels` however with 7.0,
we should modify it and the default value should be `resetchannels` for better security posture.
Also, with selectors in ACL, a client doesn't have to set channel rules everytime and by default
the value will be `resetchannels`.
Before this change
```
127.0.0.1:6379> acl list
1) "user default on nopass ~* &* +@all"
127.0.0.1:6379> acl setuser hp on nopass +@all ~*
OK
127.0.0.1:6379> acl list
1) "user default on nopass ~* &* +@all"
2) "user hp on nopass ~* &* +@all"
127.0.0.1:6379> acl setuser hp1 on nopass -@all (%R~sales*)
OK
127.0.0.1:6379> acl list
1) "user default on nopass ~* &* +@all"
2) "user hp on nopass ~* &* +@all"
3) "user hp1 on nopass &* -@all (%R~sales* &* -@all)"
```
After this change
```
127.0.0.1:6379> acl list
1) "user default on nopass ~* &* +@all"
127.0.0.1:6379> acl setuser hp on nopass +@all ~*
OK
127.0.0.1:6379> acl list
1) "user default on nopass ~* &* +@all"
2) "user hp on nopass ~* resetchannels +@all"
127.0.0.1:6379> acl setuser hp1 on nopass -@all (%R~sales*)
OK
127.0.0.1:6379> acl list
1) "user default on nopass ~* &* +@all"
2) "user hp on nopass ~* resetchannels +@all"
3) "user hp1 on nopass resetchannels -@all (%R~sales* resetchannels -@all)"
```
We recently removed capabilities from the first-arg feature of ACL and added a warning.
but we didn't document it.
ref: #10147 and redis/redis-doc#1761
For some complex data types, server.dirty actually counts
the number of elements that have been changed.
And in FLUSHDB or FLUSHALL, we count the number of keys.
So the word "key" is not strictly correct and is outdated.
Some discussion can be seen at #8140.
* Implemented selectors which provide multiple different sets of permissions to users
* Implemented key based permissions
* Added a new ACL dry-run command to test permissions before execution
* Updated module APIs to support checking key based permissions
Co-authored-by: Oran Agra <oran@redislabs.com>
Some modules might perform a long-running logic in different stages of Redis lifetime, for example:
* command execution
* RDB loading
* thread safe context
During this long-running logic Redis is not responsive.
This PR offers
1. An API to process events while a busy command is running (`RM_Yield`)
2. A new flag (`ALLOW_BUSY`) to mark the commands that should be handled during busy
jobs which can also be used by modules (`allow-busy`)
3. In slow commands and thread safe contexts, this flag will start rejecting commands with -BUSY only
after `busy-reply-threshold`
4. During loading (`rdb_load` callback), it'll process events right away (not wait for `busy-reply-threshold`),
but either way, the processing is throttled to the server hz rate.
5. Allow modules to Yield to redis background tasks, but not to client commands
* rename `script-time-limit` to `busy-reply-threshold` (an alias to the pre-7.0 `lua-time-limit`)
Co-authored-by: Oran Agra <oran@redislabs.com>
1. enable diskless replication by default
2. add a new config named repl-diskless-sync-max-replicas that enables
replication to start before the full repl-diskless-sync-delay was
reached.
3. put replica online sooner on the master (see below)
4. test suite uses repl-diskless-sync-delay of 0 to be faster
5. a few tests that use multiple replica on a pre-populated master, are
now using the new repl-diskless-sync-max-replicas
6. fix possible timing issues in a few cluster tests (see below)
put replica online sooner on the master
----------------------------------------------------
there were two tests that failed because they needed for the master to
realize that the replica is online, but the test code was actually only
waiting for the replica to realize it's online, and in diskless it could
have been before the master realized it.
changes include two things:
1. the tests wait on the right thing
2. issues in the master, putting the replica online in two steps.
the master used to put the replica as online in 2 steps. the first
step was to mark it as online, and the second step was to enable the
write event (only after getting ACK), but in fact the first step didn't
contains some of the tasks to put it online (like updating good slave
count, and sending the module event). this meant that if a test was
waiting to see that the replica is online form the point of view of the
master, and then confirm that the module got an event, or that the
master has enough good replicas, it could fail due to timing issues.
so now the full effect of putting the replica online, happens at once,
and only the part about enabling the writes is delayed till the ACK.
fix cluster tests
--------------------
I added some code to wait for the replica to sync and avoid race
conditions.
later realized the sentinel and cluster tests where using the original 5
seconds delay, so changed it to 0.
this means the other changes are probably not needed, but i suppose
they're still better (avoid race conditions)
# Short description
The Redis extended latency stats track per command latencies and enables:
- exporting the per-command percentile distribution via the `INFO LATENCYSTATS` command.
**( percentile distribution is not mergeable between cluster nodes ).**
- exporting the per-command cumulative latency distributions via the `LATENCY HISTOGRAM` command.
Using the cumulative distribution of latencies we can merge several stats from different cluster nodes
to calculate aggregate metrics .
By default, the extended latency monitoring is enabled since the overhead of keeping track of the
command latency is very small.
If you don't want to track extended latency metrics, you can easily disable it at runtime using the command:
- `CONFIG SET latency-tracking no`
By default, the exported latency percentiles are the p50, p99, and p999.
You can alter them at runtime using the command:
- `CONFIG SET latency-tracking-info-percentiles "0.0 50.0 100.0"`
## Some details:
- The total size per histogram should sit around 40 KiB. We only allocate those 40KiB when a command
was called for the first time.
- With regards to the WRITE overhead As seen below, there is no measurable overhead on the achievable
ops/sec or full latency spectrum on the client. Including also the measured redis-benchmark for unstable
vs this branch.
- We track from 1 nanosecond to 1 second ( everything above 1 second is considered +Inf )
## `INFO LATENCYSTATS` exposition format
- Format: `latency_percentiles_usec_<CMDNAME>:p0=XX,p50....`
## `LATENCY HISTOGRAM [command ...]` exposition format
Return a cumulative distribution of latencies in the format of a histogram for the specified command names.
The histogram is composed of a map of time buckets:
- Each representing a latency range, between 1 nanosecond and roughly 1 second.
- Each bucket covers twice the previous bucket's range.
- Empty buckets are not printed.
- Everything above 1 sec is considered +Inf.
- At max there will be log2(1000000000)=30 buckets
We reply a map for each command in the format:
`<command name> : { `calls`: <total command calls> , `histogram` : { <bucket 1> : latency , < bucket 2> : latency, ... } }`
Co-authored-by: Oran Agra <oran@redislabs.com>
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 implements a sharded pubsub implementation based off of shard channels.
Co-authored-by: Harkrishn Patro <harkrisp@amazon.com>
Co-authored-by: Madelyn Olson <madelyneolson@gmail.com>
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>
Block sensitive configs and commands by default.
* `enable-protected-configs` - block modification of configs with the new `PROTECTED_CONFIG` flag.
Currently we add this flag to `dbfilename`, and `dir` configs,
all of which are non-mutable configs that can set a file redis will write to.
* `enable-debug-command` - block the `DEBUG` command
* `enable-module-command` - block the `MODULE` command
These have a default value set to `no`, so that these features are not
exposed by default to client connections, and can only be set by modifying the config file.
Users can change each of these to either `yes` (allow all access), or `local` (allow access from
local TCP connections and unix domain connections)
Note that this is a **breaking change** (specifically the part about MODULE command being disabled by default).
I.e. we don't consider DEBUG command being blocked as an issue (people shouldn't have been using it),
and the few configs we protected are unlikely to have been set at runtime anyway.
On the other hand, it's likely to assume some users who use modules, load them from the config file anyway.
Note that's the whole point of this PR, for redis to be more secure by default and reduce the attack surface on
innocent users, so secure defaults will necessarily mean a breaking change.
Introduce memory management on cluster link buffers:
* Introduce a new `cluster-link-sendbuf-limit` config that caps memory usage of cluster bus link send buffers.
* Introduce a new `CLUSTER LINKS` command that displays current TCP links to/from peers.
* Introduce a new `mem_cluster_links` field under `INFO` command output, which displays the overall memory usage by all current cluster links.
* Introduce a new `total_cluster_links_buffer_limit_exceeded` field under `CLUSTER INFO` command output, which displays the accumulated count of cluster links freed due to `cluster-link-sendbuf-limit`.
We can now do: `config set maxmemory 10m repl-backlog-size 5m`
## Basic algorithm to support "transaction like" config sets:
1. Backup all relevant current values (via get).
2. Run "verify" and "set" on everything, if we fail run "restore".
3. Run "apply" on everything (optional optimization: skip functions already run). If we fail run "restore".
4. Return success.
### restore
1. Run set on everything in backup. If we fail log it and continue (this puts us in an undefined
state but we decided it's better than the alternative of panicking). This indicates either a bug
or some unsupported external state.
2. Run apply on everything in backup (optimization: skip functions already run). If we fail log
it (see comment above).
3. Return error.
## Implementation/design changes:
* Apply function are idempotent (have no effect if they are run more than once for the same config).
* No indication in set functions if we're reading the config or running from the `CONFIG SET` command
(removed `update` argument).
* Set function should set some config variable and assume an (optional) apply function will use that
later to apply. If we know this setting can be safely applied immediately and can always be reverted
and doesn't depend on any other configuration we can apply immediately from within the set function
(and not store the setting anywhere). This is the case of this `dir` config, for example, which has no
apply function. No apply function is need also in the case that setting the variable in the `server` struct
is all that needs to be done to make the configuration take effect. Note that the original concept of `update_fn`,
which received the old and new values was removed and replaced by the optional apply function.
* Apply functions use settings written to the `server` struct and don't receive any inputs.
* I take care that for the generic (non-special) configs if there's no change I avoid calling the setter (possible
optimization: avoid calling the apply function as well).
* Passing the same config parameter more than once to `config set` will fail. You can't do `config set my-setting
value1 my-setting value2`.
Note that getting `save` in the context of the conf file parsing to work here as before was a pain.
The conf file supports an aggregate `save` definition, where each `save` line is added to the server's
save params. This is unlike any other line in the config file where each line overwrites any previous
configuration. Since we now support passing multiple save params in a single line (see top comments
about `save` in https://github.com/redis/redis/pull/9644) we should deprecate the aggregate nature of
this config line and perhaps reduce this ugly code in the future.
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>
Currently PING returns different status when server is not serving data,
for example when `LOADING` or `BUSY`.
But same was not true for `MASTERDOWN`
This commit makes PING reply with `MASTERDOWN` when
replica-serve-stale-data=no and link is MASTER is down.
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>
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.
Yuan Wang