upstream/redis
1
0
Fork 0
mirror of https://github.com/redis/redis.git synced 2026-02-03 20:39:54 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions 3
redis/src/multi.c
Yuan Wang 64a40b20d9
Async IO Threads (#13695) 
## 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:
redis#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>
2024-12-23 14:16:40 +08:00

488 lines
17 KiB
C
Raw Blame History

/*
* Copyright (c) 2009-Present, Redis Ltd.
* All rights reserved.
*
* Licensed under your choice of the Redis Source Available License 2.0
* (RSALv2) or the Server Side Public License v1 (SSPLv1).
*/
#include "server.h"
/* ================================ MULTI/EXEC ============================== */
/* Client state initialization for MULTI/EXEC */
void initClientMultiState(client *c) {
c->mstate.commands = NULL;
c->mstate.count = 0;
c->mstate.cmd_flags = 0;
c->mstate.cmd_inv_flags = 0;
c->mstate.argv_len_sums = 0;
c->mstate.alloc_count = 0;
}
/* Release all the resources associated with MULTI/EXEC state */
void freeClientMultiState(client *c) {
int j;
for (j = 0; j < c->mstate.count; j++) {
int i;
multiCmd *mc = c->mstate.commands+j;
for (i = 0; i < mc->argc; i++)
decrRefCount(mc->argv[i]);
zfree(mc->argv);
}
zfree(c->mstate.commands);
}
/* Add a new command into the MULTI commands queue */
void queueMultiCommand(client *c, uint64_t cmd_flags) {
multiCmd *mc;
/* No sense to waste memory if the transaction is already aborted.
* this is useful in case client sends these in a pipeline, or doesn't
* bother to read previous responses and didn't notice the multi was already
* aborted. */
if (c->flags & (CLIENT_DIRTY_CAS|CLIENT_DIRTY_EXEC))
return;
if (c->mstate.count == 0) {
/* If a client is using multi/exec, assuming it is used to execute at least
* two commands. Hence, creating by default size of 2. */
c->mstate.commands = zmalloc(sizeof(multiCmd)*2);
c->mstate.alloc_count = 2;
}
if (c->mstate.count == c->mstate.alloc_count) {
c->mstate.alloc_count = c->mstate.alloc_count < INT_MAX/2 ? c->mstate.alloc_count*2 : INT_MAX;
c->mstate.commands = zrealloc(c->mstate.commands, sizeof(multiCmd)*(c->mstate.alloc_count));
}
mc = c->mstate.commands+c->mstate.count;
mc->cmd = c->cmd;
mc->argc = c->argc;
mc->argv = c->argv;
mc->argv_len = c->argv_len;
c->mstate.count++;
c->mstate.cmd_flags |= cmd_flags;
c->mstate.cmd_inv_flags |= ~cmd_flags;
c->mstate.argv_len_sums += c->argv_len_sum + sizeof(robj*)*c->argc;
/* Reset the client's args since we copied them into the mstate and shouldn't
* reference them from c anymore. */
c->argv = NULL;
c->argc = 0;
c->argv_len_sum = 0;
c->argv_len = 0;
}
void discardTransaction(client *c) {
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(CLIENT_MULTI|CLIENT_DIRTY_CAS|CLIENT_DIRTY_EXEC);
unwatchAllKeys(c);
}
/* Flag the transaction as DIRTY_EXEC so that EXEC will fail.
* Should be called every time there is an error while queueing a command. */
void flagTransaction(client *c) {
if (c->flags & CLIENT_MULTI)
c->flags |= CLIENT_DIRTY_EXEC;
}
void multiCommand(client *c) {
if (c->flags & CLIENT_MULTI) {
addReplyError(c,"MULTI calls can not be nested");
return;
}
c->flags |= CLIENT_MULTI;
addReply(c,shared.ok);
}
void discardCommand(client *c) {
if (!(c->flags & CLIENT_MULTI)) {
addReplyError(c,"DISCARD without MULTI");
return;
}
discardTransaction(c);
addReply(c,shared.ok);
}
/* Aborts a transaction, with a specific error message.
* The transaction is always aborted with -EXECABORT so that the client knows
* the server exited the multi state, but the actual reason for the abort is
* included too.
* Note: 'error' may or may not end with \r\n. see addReplyErrorFormat. */
void execCommandAbort(client *c, sds error) {
discardTransaction(c);
if (error[0] == '-') error++;
addReplyErrorFormat(c, "-EXECABORT Transaction discarded because of: %s", error);
/* Send EXEC to clients waiting data from MONITOR. We did send a MULTI
* already, and didn't send any of the queued commands, now we'll just send
* EXEC so it is clear that the transaction is over. */
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
void execCommand(client *c) {
int j;
robj **orig_argv;
int orig_argc, orig_argv_len;
struct redisCommand *orig_cmd;
if (!(c->flags & CLIENT_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
/* EXEC with expired watched key is disallowed*/
if (isWatchedKeyExpired(c)) {
c->flags |= (CLIENT_DIRTY_CAS);
}
/* Check if we need to abort the EXEC because:
* 1) Some WATCHed key was touched.
* 2) There was a previous error while queueing commands.
* A failed EXEC in the first case returns a multi bulk nil object
* (technically it is not an error but a special behavior), while
* in the second an EXECABORT error is returned. */
if (c->flags & (CLIENT_DIRTY_CAS | CLIENT_DIRTY_EXEC)) {
if (c->flags & CLIENT_DIRTY_EXEC) {
addReplyErrorObject(c, shared.execaborterr);
} else {
addReply(c, shared.nullarray[c->resp]);
}
discardTransaction(c);
return;
}
uint64_t old_flags = c->flags;
/* we do not want to allow blocking commands inside multi */
c->flags |= CLIENT_DENY_BLOCKING;
/* Exec all the queued commands */
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
server.in_exec = 1;
orig_argv = c->argv;
orig_argv_len = c->argv_len;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyArrayLen(c,c->mstate.count);
for (j = 0; j < c->mstate.count; j++) {
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->argv_len = c->mstate.commands[j].argv_len;
c->cmd = c->realcmd = c->mstate.commands[j].cmd;
/* ACL permissions are also checked at the time of execution in case
* they were changed after the commands were queued. */
int acl_errpos;
int acl_retval = ACLCheckAllPerm(c,&acl_errpos);
if (acl_retval != ACL_OK) {
char *reason;
switch (acl_retval) {
case ACL_DENIED_CMD:
reason = "no permission to execute the command or subcommand";
break;
case ACL_DENIED_KEY:
reason = "no permission to touch the specified keys";
break;
case ACL_DENIED_CHANNEL:
reason = "no permission to access one of the channels used "
"as arguments";
break;
default:
reason = "no permission";
break;
}
addACLLogEntry(c,acl_retval,ACL_LOG_CTX_MULTI,acl_errpos,NULL,NULL);
addReplyErrorFormat(c,
"-NOPERM ACLs rules changed between the moment the "
"transaction was accumulated and the EXEC call. "
"This command is no longer allowed for the "
"following reason: %s", reason);
} else {
if (c->id == CLIENT_ID_AOF)
call(c,CMD_CALL_NONE);
else
call(c,CMD_CALL_FULL);
serverAssert((c->flags & CLIENT_BLOCKED) == 0);
}
/* Commands may alter argc/argv, restore mstate. */
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].argv_len = c->argv_len;
c->mstate.commands[j].cmd = c->cmd;
}
// restore old DENY_BLOCKING value
if (!(old_flags & CLIENT_DENY_BLOCKING))
c->flags &= ~CLIENT_DENY_BLOCKING;
c->argv = orig_argv;
c->argv_len = orig_argv_len;
c->argc = orig_argc;
c->cmd = c->realcmd = orig_cmd;
discardTransaction(c);
server.in_exec = 0;
}
/* ===================== WATCH (CAS alike for MULTI/EXEC) ===================
*
* The implementation uses a per-DB hash table mapping keys to list of clients
* WATCHing those keys, so that given a key that is going to be modified
* we can mark all the associated clients as dirty.
*
* Also every client contains a list of WATCHed keys so that's possible to
* un-watch such keys when the client is freed or when UNWATCH is called. */
/* The watchedKey struct is included in two lists: the client->watched_keys list,
* and db->watched_keys dict (each value in that dict is a list of watchedKey structs).
* The list in the client struct is a plain list, where each node's value is a pointer to a watchedKey.
* The list in the db db->watched_keys is different, the listnode member that's embedded in this struct
* is the node in the dict. And the value inside that listnode is a pointer to the that list, and we can use
* struct member offset math to get from the listnode to the watchedKey struct.
* This is done to avoid the need for listSearchKey and dictFind when we remove from the list. */
typedef struct watchedKey {
listNode node;
robj *key;
redisDb *db;
client *client;
unsigned expired:1; /* Flag that we're watching an already expired key. */
} watchedKey;
/* Attach a watchedKey to the list of clients watching that key. */
static inline void watchedKeyLinkToClients(list *clients, watchedKey *wk) {
wk->node.value = clients; /* Point the value back to the list */
listLinkNodeTail(clients, &wk->node); /* Link the embedded node */
}
/* Get the list of clients watching that key. */
static inline list *watchedKeyGetClients(watchedKey *wk) {
return listNodeValue(&wk->node); /* embedded node->value points back to the list */
}
/* Get the node with wk->client in the list of clients watching that key. Actually it
* is just the embedded node. */
static inline listNode *watchedKeyGetClientNode(watchedKey *wk) {
return &wk->node;
}
/* Watch for the specified key */
void watchForKey(client *c, robj *key) {
list *clients = NULL;
listIter li;
listNode *ln;
watchedKey *wk;
if (listLength(c->watched_keys) == 0) server.watching_clients++;
/* Check if we are already watching for this key */
listRewind(c->watched_keys,&li);
while((ln = listNext(&li))) {
wk = listNodeValue(ln);
if (wk->db == c->db && equalStringObjects(key,wk->key))
return; /* Key already watched */
}
/* This key is not already watched in this DB. Let's add it */
clients = dictFetchValue(c->db->watched_keys,key);
if (!clients) {
clients = listCreate();
dictAdd(c->db->watched_keys,key,clients);
incrRefCount(key);
}
/* Add the new key to the list of keys watched by this client */
wk = zmalloc(sizeof(*wk));
wk->key = key;
wk->client = c;
wk->db = c->db;
wk->expired = keyIsExpired(c->db, key);
incrRefCount(key);
listAddNodeTail(c->watched_keys, wk);
watchedKeyLinkToClients(clients, wk);
}
/* Unwatch all the keys watched by this client. To clean the EXEC dirty
* flag is up to the caller. */
void unwatchAllKeys(client *c) {
listIter li;
listNode *ln;
if (listLength(c->watched_keys) == 0) return;
listRewind(c->watched_keys,&li);
while((ln = listNext(&li))) {
list *clients;
watchedKey *wk;
/* Remove the client's wk from the list of clients watching the key. */
wk = listNodeValue(ln);
clients = watchedKeyGetClients(wk);
serverAssertWithInfo(c,NULL,clients != NULL);
listUnlinkNode(clients, watchedKeyGetClientNode(wk));
/* Kill the entry at all if this was the only client */
if (listLength(clients) == 0)
dictDelete(wk->db->watched_keys, wk->key);
/* Remove this watched key from the client->watched list */
listDelNode(c->watched_keys,ln);
decrRefCount(wk->key);
zfree(wk);
}
server.watching_clients--;
}
/* Iterates over the watched_keys list and looks for an expired key. Keys which
* were expired already when WATCH was called are ignored. */
int isWatchedKeyExpired(client *c) {
listIter li;
listNode *ln;
watchedKey *wk;
if (listLength(c->watched_keys) == 0) return 0;
listRewind(c->watched_keys,&li);
while ((ln = listNext(&li))) {
wk = listNodeValue(ln);
if (wk->expired) continue; /* was expired when WATCH was called */
if (keyIsExpired(wk->db, wk->key)) return 1;
}
return 0;
}
/* "Touch" a key, so that if this key is being WATCHed by some client the
* next EXEC will fail.
*
* Sanitizer suppression: IO threads also read c->flags, but never modify
* it or read the CLIENT_DIRTY_CAS bit, main thread just only modifies
* this bit, so there is actually no real data race. */
REDIS_NO_SANITIZE("thread")
void touchWatchedKey(redisDb *db, robj *key) {
list *clients;
listIter li;
listNode *ln;
if (dictSize(db->watched_keys) == 0) return;
clients = dictFetchValue(db->watched_keys, key);
if (!clients) return;
/* Mark all the clients watching this key as CLIENT_DIRTY_CAS */
/* Check if we are already watching for this key */
listRewind(clients,&li);
while((ln = listNext(&li))) {
watchedKey *wk = redis_member2struct(watchedKey, node, ln);
client *c = wk->client;
if (wk->expired) {
/* The key was already expired when WATCH was called. */
if (db == wk->db &&
equalStringObjects(key, wk->key) &&
dbFind(db, key->ptr) == NULL)
{
/* Already expired key is deleted, so logically no change. Clear
* the flag. Deleted keys are not flagged as expired. */
wk->expired = 0;
goto skip_client;
}
break;
}
c->flags |= CLIENT_DIRTY_CAS;
/* As the client is marked as dirty, there is no point in getting here
* again in case that key (or others) are modified again (or keep the
* memory overhead till EXEC). */
unwatchAllKeys(c);
skip_client:
continue;
}
}
/* Set CLIENT_DIRTY_CAS to all clients of DB when DB is dirty.
* It may happen in the following situations:
* FLUSHDB, FLUSHALL, SWAPDB, end of successful diskless replication.
*
* replaced_with: for SWAPDB, the WATCH should be invalidated if
* the key exists in either of them, and skipped only if it
* doesn't exist in both. */
REDIS_NO_SANITIZE("thread")
void touchAllWatchedKeysInDb(redisDb *emptied, redisDb *replaced_with) {
listIter li;
listNode *ln;
dictEntry *de;
if (dictSize(emptied->watched_keys) == 0) return;
dictIterator *di = dictGetSafeIterator(emptied->watched_keys);
while((de = dictNext(di)) != NULL) {
robj *key = dictGetKey(de);
int exists_in_emptied = dbFind(emptied, key->ptr) != NULL;
if (exists_in_emptied ||
(replaced_with && dbFind(replaced_with, key->ptr) != NULL))
{
list *clients = dictGetVal(de);
if (!clients) continue;
listRewind(clients,&li);
while((ln = listNext(&li))) {
watchedKey *wk = redis_member2struct(watchedKey, node, ln);
if (wk->expired) {
if (!replaced_with || !dbFind(replaced_with, key->ptr)) {
/* Expired key now deleted. No logical change. Clear the
* flag. Deleted keys are not flagged as expired. */
wk->expired = 0;
continue;
} else if (keyIsExpired(replaced_with, key)) {
/* Expired key remains expired. */
continue;
}
} else if (!exists_in_emptied && keyIsExpired(replaced_with, key)) {
/* Non-existing key is replaced with an expired key. */
wk->expired = 1;
continue;
}
client *c = wk->client;
c->flags |= CLIENT_DIRTY_CAS;
/* Note - we could potentially call unwatchAllKeys for this specific client in order to reduce
* the total number of iterations. BUT this could also free the current next entry pointer
* held by the iterator and can lead to use-after-free. */
}
}
}
dictReleaseIterator(di);
}
void watchCommand(client *c) {
int j;
if (c->flags & CLIENT_MULTI) {
addReplyError(c,"WATCH inside MULTI is not allowed");
return;
}
/* No point in watching if the client is already dirty. */
if (c->flags & CLIENT_DIRTY_CAS) {
addReply(c,shared.ok);
return;
}
for (j = 1; j < c->argc; j++)
watchForKey(c,c->argv[j]);
addReply(c,shared.ok);
}
void unwatchCommand(client *c) {
unwatchAllKeys(c);
c->flags &= (~CLIENT_DIRTY_CAS);
addReply(c,shared.ok);
}
size_t multiStateMemOverhead(client *c) {
size_t mem = c->mstate.argv_len_sums;
/* Add watched keys overhead, Note: this doesn't take into account the watched keys themselves, because they aren't managed per-client. */
mem += listLength(c->watched_keys) * (sizeof(listNode) + sizeof(watchedKey));
/* Reserved memory for queued multi commands. */
mem += c->mstate.alloc_count * sizeof(multiCmd);
return mem;
}
Reference in a new issue View git blame Copy permalink