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outdated: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
We saw some tests sporadically time out on valgrind (namely the ones
from #9323).
Increasing valgrind timeout from 20 mins to 40 mins in CI.
And fixing an outdated help message.
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.
During diskless replication, the check for broken EOF mark is misplaced
and should be earlier. Now we do not swap db, we do proper cleanup and
correctly raise module events on this kind of failure.
This issue existed prior to #9323, but before, the side effect was not restoring
backup and not raising the correct module events on this failure.
* Clean up EINTR handling so EINTR will not change connection state to begin with.
* On TLS, catch EINTR and return it as-is before going through OpenSSL error handling (which seems to not distinguish it from EAGAIN).
Optimized port detection for tcl, use 'socket -server' instead of 'socket' to rule out port on TIME_WAIT
Co-authored-by: chendianqiang <chendianqiang@meituan.com>
Co-authored-by: Oran Agra <oran@redislabs.com>
This refactors all `CONFIG SET`s and conf file loading arguments go through
the generic config handling interface.
Refactoring changes:
- All config params go through the `standardConfig` interface (some stuff which
is only related to the config file and not the `CONFIG` command still has special
handling for rewrite/config file parsing, `loadmodule`, for example.) .
- Added `MULTI_ARG_CONFIG` flag for configs to signify they receive a variable
number of arguments instead of a single argument. This is used to break up space
separated arguments to `CONFIG SET` so the generic setter interface can pass
multiple arguments to the setter function. When parsing the config file we also break
up anything after the config name into multiple arguments to the setter function.
Interface changes:
- A side effect of the above interface is that the `bind` argument in the config file can
be empty (no argument at all) this is treated the same as passing an single empty
string argument (same as `save` already used to work).
- Support rewrite and setting `watchdog-period` from config file (was only supported
by the CONFIG command till now).
- Another side effect is that the `save T X` config argument now supports multiple
Time-Changes pairs in a single line like its `CONFIG SET` counterpart. So in the
config file you can either do:
```
save 3600 1
save 600 10
```
or do
```
save 3600 1 600 10
```
Co-authored-by: Bjorn Svensson <bjorn.a.svensson@est.tech>
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>
Introduced in #8179, this fixes the command's replies in the 0 count edge case.
[BREAKING] changes the reply type when count is 0 to an empty array (instead of nil)
Moves LPOP ... 0 fast exit path after type check to reply with WRONGTYPE
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>
there was a chance that by the time the assertion is executed,
the replica already manages to reconnect.
now we make sure the replica is unable to re-connect to the master.
additionally, we wait for some gossip from the disconnected replica,
to see that it doesn't mess things up.
unrelated: fix a typo when trying to exhaust the backlog, one that
didn't have any harmful implications
Co-authored-by: Madelyn Olson <madelyneolson@gmail.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>
Add new no-mandatory-keys flag to support COMMAND GETKEYS of commands
which have no mandatory keys.
In the past we would have got this error:
```
127.0.0.1:6379> command getkeys eval "return 1" 0
(error) ERR Invalid arguments specified for command
```
When using SETNX and SETXX we could end up doing key lookup twice.
This presents a small inefficiency price.
Also once we have statistics of write hit and miss they'll be wrong (recording the same key hit twice)
The issue was that setting maxmemory to used_memory and expecting
eviction is insufficient, since we need to take
mem_not_counted_for_evict into consideration.
This test got broken by #9166
The External tests started failing recently for unclear reason:
```
*** [err]: Tracking invalidation message of eviction keys should be before response in tests/unit/tracking.tcl
Expected '0' to be equal to 'invalidate volatile-key' (context: type eval line 21 cmd {assert_equal $res {invalidate volatile-key}} proc ::test)
```
I suspect the issue is that the used_memory sample is taken while a lazy free is still being processed.
Since the loop in incrementalTrimReplicationBacklog checks the size of histlen,
we cannot afford to update it only when the loop exits, this may cause deleting
much more replication blocks, and replication backlog may be less than setting size.
introduce in #9166
Co-authored-by: sundb <sundbcn@gmail.com>
After PR #9166 , replication backlog is not a real block of memory, just contains a
reference points to replication buffer's block and the blocks index (to accelerate
search offset when partial sync), so we need update both replication buffer's block's
offset and replication backlog blocks index's offset when master restart from RDB,
since the `server.master_repl_offset` is changed.
The implications of this bug was just a slow search, but not a replication failure.
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>
The module test in reply.tcl was introduced by #8521 but didn't run until recently (see #9639)
and then it started failing with valgrind.
This is because valgrind uses 64 bit long double (unlike most other platforms that have at least 80 bits)
But besides valgrind, the tests where also incompatible with ARM32, which also uses 64 bit long doubles.
We now use appropriate value to avoid issues with either valgrind or ARM32
In all the double tests, i use 3.141, which is safe since since addReplyDouble uses
`%.17Lg` which is able to represent this value without adding any digits due to precision loss.
In the long double, since we use `%.17Lf` in ld2string, it preserves 17 significant
digits, rather than 17 digit after the decimal point (like in `%.17Lg`).
So to make these similar, i use value lower than 1 (no digits left of
the period)
Lastly, we have the same issue with TCL (no long doubles) so we read
raw protocol in that test.
Note that the only error before this fix (in both valgrind and ARM32 is this:
```
*** [err]: RM_ReplyWithLongDouble: a float reply in tests/unit/moduleapi/reply.tcl
Expected '3.141' to be equal to '3.14100000000000001' (context: type eval line 2 cmd {assert_equal 3.141 [r rw.longdouble 3.141]} proc ::test)
```
so the changes to debug.c and scripting.tcl aren't really needed, but i consider them a cleanup
(i.e. scripting.c validated a different constant than the one that's sent to it from debug.c).
Another unrelated change is to add the RESP version to the repeated tests in reply.tcl
Fix failures introduced by #9695 which was an attempt to solve failures introduced by #9679.
And alternative to #9703 (i didn't like the extra argument to kill_instance).
Reverting #9695.
Instead of stopping AOF on all terminations, stop it only on the two which need it.
Do it as part of the test rather than the infra (it was add that kill_instance used `R`
to communicate to the instance)
Note that the original purpose of these tests was to trigger a crash, but that upsets
valgrind so in redis 6.2 i changed it to use SIGTERM, so i now rename the tests
(remove "kill" and "crash").
Also add some colors to failures, and the word "FAILED" so that it's searchable.
And solve a semi-related race condition in 14-consistency-check.tcl
This solves several problems in a more elegant way:
* No need to explicitly use `-lc` on x86_64 when building with `-m32`.
* Avoids issues with undefined floating point emulation funcs on ARM.
The previous code did not check whether COUNT is set.
So we can use `lmpop 2 key1 key2 left count 1 count 2`.
This situation can occur in LMPOP/BLMPOP/ZMPOP/BZMPOP commands.
LMPOP/BLMPOP introduced in #9373, ZMPOP/BZMPOP introduced in #9484.
When stopping an instance in the cluster tests, disable appendonly first, so that SIGTERM won't be ignored.
Recently in #9679 i change the test infra to use SIGSEGV to kill servers that refuse
the SIGTERM rather than do SIGKILL directly.
This surfaced an issue that i've added in #7725 which changed SIGKILL to SIGTERM (to resolve valgrind issues).
So the current situation in the past months was that sometimes servers refused the
SIGTERM and waited 10 seconds for the SIGKILL, and this commit resolves that (faster termination).
I recently started seeing a lot of empty valgrind reports in the daily CI.
i.e. prints showing valgrind header but no leak report, which causes the tests to fail
https://github.com/redis/redis/runs/3991335416?check_suite_focus=true
This commit change 2 things:
* first, considering valgrind is just slow, we used to give processes 60 seconds timeout on shutdown
instead of 10 seconds we give normally. this commit changes that to 120.
* secondly, when we reach the timeout, we first try to use SIGSEGV so that maybe we'll get a stack
trace indicating where redis is hang, and we only resort to SIGKILL if double that time passed.
note that if there are indeed hang processes, we will normally not see that in the non-valgrind runs,
since the tests didn't use to detect any failure in that case, and now they will since `crashlog_from_file`
is run after `kill_server`.
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)
Improve code doc for allowed_firstargs (used to be allowed_commands before #9504.
I don't think the text in the code needs to refer to the history (it's not there just for backwards compatibility).
instead it should just describe what it does.
Let modules use additional type of RESP3 response (unused by redis so far)
Also fix tests that where introduced in #8521 but didn't actually run.
Co-authored-by: Oran Agra <oran@redislabs.com>
## 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.
I moved a bunch of stats in redisFork to be executed only on successful
fork, since they seem wrong to be done when it failed.
I guess when fork fails it does that immediately, no latency spike.
