* need to include stdint for uintptr_t
* need to include stdint for uintptr_t
* use atomic_load for g_pserver->mstime
* use atomic_load for g_pserver->mstime
* Integrate readwritelock with Pro Code
* Integrate readwritelock with Pro Code
* Defensive asserts for RWLock
* Defensive asserts for RWLock
* Save and restore master info in rdb to allow active replica partial sync (#371)
* save replid for all masters in rdb
* expanded rdbSaveInfo to hold multiple master structs
* parse repl-masters from rdb
* recover replid info from rdb in active replica mode, attempt partial sync
* save offset from rdb into correct variable
* don't change replid based on master in active rep
* save and load psync info from correct fields
* Save and restore master info in rdb to allow active replica partial sync (#371)
* save replid for all masters in rdb
* expanded rdbSaveInfo to hold multiple master structs
* parse repl-masters from rdb
* recover replid info from rdb in active replica mode, attempt partial sync
* save offset from rdb into correct variable
* don't change replid based on master in active rep
* save and load psync info from correct fields
* placement new instead of memcpy
* placement new instead of memcpy
* Remove asserts, RW lock can go below zero in cases of aeAcquireLock
* Remove asserts, RW lock can go below zero in cases of aeAcquireLock
* Inclusive language
* Inclusive language
* update packaging for OS merge
* update packaging for OS merge
* modify dockerfile to build within image
* modify dockerfile to build within image
* Make active client balancing a configurable option
* Make active client balancing a configurable option
* With TLS throttle accepts if server is under heavy load - do not change non TLS behavior
* With TLS throttle accepts if server is under heavy load - do not change non TLS behavior
* Only run the tls-name-validation test if --tls is passed into runtest
* Only run the tls-name-validation test if --tls is passed into runtest
* Fix KeyDB not building with TLS < 1.1.1
* Fix KeyDB not building with TLS < 1.1.1
* update changelog to use replica as terminology
* update changelog to use replica as terminology
* update copyright
* update copyright
* update deb copyright
* update deb copyright
* call aeThreadOnline() earlier
* call aeThreadOnline() earlier
* Removed mergeReplicationId
* Removed mergeReplicationId
* acceptTLS is threadsafe like the non TLS version
* acceptTLS is threadsafe like the non TLS version
* setup Machamp ci
* setup Machamp ci
* make build_test.sh executable
* make build_test.sh executable
* PSYNC production fixes
* PSYNC production fixes
* fix the Machamp build
* fix the Machamp build
* break into tests into steps
* break into tests into steps
* Added multimaster test
* Added multimaster test
* Update ci.yml
Change min tested version to 18.04
* Update ci.yml
Change min tested version to 18.04
* fork lock for all threads, use fastlock for readwritelock
* fork lock for all threads, use fastlock for readwritelock
* hide forklock object in ae
* hide forklock object in ae
* only need to include readwritelock in ae
* only need to include readwritelock in ae
* time thread lock uses fastlock instead of std::mutex
* time thread lock uses fastlock instead of std::mutex
* set thread as offline when waiting for time thread lock
* set thread as offline when waiting for time thread lock
* update README resource links
* update README resource links
* Fix MALLOC=memkind build issues
* Fix MALLOC=memkind build issues
* Fix module test break
* Fix module test break
* Eliminate firewall dialogs on mac for regular and cluster tests. There are still issues with the sentinel tests but attempting to bind only to localhost causes failures
* Eliminate firewall dialogs on mac for regular and cluster tests. There are still issues with the sentinel tests but attempting to bind only to localhost causes failures
* remove unused var in networking.cpp
* remove unused var in networking.cpp
* check ziplist len to avoid crash on empty ziplist convert
* check ziplist len to avoid crash on empty ziplist convert
* remove nullptr subtraction
* remove nullptr subtraction
* cannot mod a pointer
* cannot mod a pointer
* need to include stdint for uintptr_t
* need to include stdint for uintptr_t
* use atomic_load for g_pserver->mstime
* use atomic_load for g_pserver->mstime
* Integrate readwritelock with Pro Code
* Integrate readwritelock with Pro Code
* Defensive asserts for RWLock
* Defensive asserts for RWLock
* Save and restore master info in rdb to allow active replica partial sync (#371)
* save replid for all masters in rdb
* expanded rdbSaveInfo to hold multiple master structs
* parse repl-masters from rdb
* recover replid info from rdb in active replica mode, attempt partial sync
* save offset from rdb into correct variable
* don't change replid based on master in active rep
* save and load psync info from correct fields
* Save and restore master info in rdb to allow active replica partial sync (#371)
* save replid for all masters in rdb
* expanded rdbSaveInfo to hold multiple master structs
* parse repl-masters from rdb
* recover replid info from rdb in active replica mode, attempt partial sync
* save offset from rdb into correct variable
* don't change replid based on master in active rep
* save and load psync info from correct fields
* placement new instead of memcpy
* placement new instead of memcpy
* Remove asserts, RW lock can go below zero in cases of aeAcquireLock
* Remove asserts, RW lock can go below zero in cases of aeAcquireLock
* Inclusive language
* Inclusive language
* call aeThreadOnline() earlier
* call aeThreadOnline() earlier
* Removed mergeReplicationId
* Removed mergeReplicationId
* Make active client balancing a configurable option
* Make active client balancing a configurable option
* With TLS throttle accepts if server is under heavy load - do not change non TLS behavior
* With TLS throttle accepts if server is under heavy load - do not change non TLS behavior
* acceptTLS is threadsafe like the non TLS version
* acceptTLS is threadsafe like the non TLS version
* PSYNC production fixes
* PSYNC production fixes
* Ensure we are responsive during storagecache clears
* Ensure we are responsive during storagecache clears
* Ensure recreated tables use the same settings as ones made at boot
* Ensure recreated tables use the same settings as ones made at boot
* Converted some existing PSYNC tests for multimaster
* Converted some existing PSYNC tests for multimaster
* Inclusive language fix
* Inclusive language fix
* Cleanup test suite
* Cleanup test suite
* Updated test replica configs so tests make sense
* Updated test replica configs so tests make sense
* active-rep test reliability
* active-rep test reliability
* Quick fix to make psync tests work
* Quick fix to make psync tests work
* Fix PSYNC test crashes
* Fix PSYNC test crashes
* Ensure we force moves not copies when ingesting bulk insert files
* Ensure we force moves not copies when ingesting bulk insert files
* Disable async for hget commands as it is not ready
* Disable FLASH
* Fix crash in save of masterinfo
* Fix musl/Alpine build failures
* Remove unnecessary libs
* update readme
* update readme
* remove Enterprise references
* Limit max overage to 20% during RDB save
* Delete COPYING to replace with BSD license
* update deb master changelog
* Update license
* Fix Readme typo from github org transition
Replace mention of scratch-file-path with db-s3-object
* Fix reference counting failure in the dict. This is caused by std::swap also swapping refcounts
* Fix assertion in async rehash
* Prevent crash on shutdown by avoiding dtors (they are unnecessary anyways)
* Initialize noshrink, it was dangling
* Prevent us from starting a rehash when one wasn't already in progress. This can cause severe issues for snapshots
* Avoid unnecessary rehashing when a rehash is abandoned
* Dictionary use correct acquire/release semantics
* Add fence barriers for the repl backlog (important for AARCH64 and other weak memory models)
* Silence TSAN errors on ustime and mstime. Every CPU we support is atomic on aligned ints, but correctness matters
* Disable async commands by default
* Fix TSAN warnings on the repl backlog
* Merge OSS back into pro
* Fix unmerged files
* Fix O(n^2) algorithm in the GC cleanup logic
* Fix crash in expire when a snapshot is in flight. Caused by a perf optimization getting the expire map out of sync with the val
* On Alpine we must have a reasonable stack size
* Revert ci.yml to unstable branch version
* Implements the soft shutdown feature to allow clients to cooperatively disconnect preventing disruption during shutdown
* Ensure clean shutdown with multiple threads
* update dockerfiles
* update deb pkg references and changelog
* update gem reference
* lpGetInteger returns int64_t, avoid overflow (#10068)
Fix#9410
Crucial for the ms and sequence deltas, but I changed all
calls, just in case (e.g. "flags")
Before this commit:
`ms_delta` and `seq_delta` could have overflown, causing `currid` to be wrong,
which in turn would cause `streamTrim` to trim the entire rax node (see new test)
* Fix issue #454 (BSD build break)
* Do not allow commands to run in background when in eval, Issue #452
* Fix certificate leak during connection when tls-allowlists are used
* Fix issue #480
* Fix crash running INFO command while a disk based backlog is set
* check tracking per db
* fix warnings
* Fix a race when undoConnectWithMaster changes mi->repl_transfer_s but the connection is not yet closed and the event handler runs
* Fix a race in processChanges/trackChanges with rdbLoadRio by acquiring the lock when trackChanges is set
* Fix ASAN use after free
* Additional fixes
* Fix integer overflow of the track changes counter
* Fix P99 latency issue for TLS where we leave work for the next event loop
tlsProcessPendingData() needs to be called before we execute queued commands because it may enqueue more commands
* Fix race removing key cache
* Prevent crash on load in long running KeyDB instances
* Fixes a crash where the server assertion failed when the key exists in DB during RDB load
* Remove old assertion which is commented out.
* avoid from instatiating EpochHolder multiple times to improve performance and cpu utilization
* avoid from instatiating EpochHolder multiple times to improve performance and cpu utilization
* src\redis-cli.c: fix potential null pointer dereference found by cppcheck
src\redis-cli.c:5488:35: warning: Either the condition
'!table' is redundant or there is possible null pointer dereference:
table. [nullPointerRedundantCheck]
* Fix Issue #486
* Workaround bug in snapshot sync - abort don't crash
* Improve reliability of async parts of the soft shutdown tests
* Improve reliability of fragmentation tests
* Verify that partial syncs do indeed occur
* Fix O(n) algorithm in INFO command
* Remove incorrect assert that fires when the repl backlog is used fully
* Make building flash optional
* Remove unneeded gitlab CI file
* [BUG] Moves key to another DB, the source key was removed if the move failed due to the key exists in the destination db #497 (#498)
Co-authored-by: Paul Chen <mingchen@Mings-MacBook-Pro.local>
* trigger repl_curr_off!= master_repl_offset assert failure when having pending write case
* use debug for logging the message instead
* rocksdb log using up the diskspace on flash (#519)
* Fix OpenSSL 3.0.x related issues. (#10291)
* 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.
* remove unnecessary forward declaration
* remove internal ci stuff
* remove more internal ci/publishing
* submodule update step
* use with syntax instead
* bump ci ubuntu old ver as latest is now 22.04
* include submodules on all ci jobs
* install all deps for all ci jobs
Co-authored-by: Vivek Saini <vsaini@snapchat.com>
Co-authored-by: Christian Legge <christian@eqalpha.com>
Co-authored-by: benschermel <bschermel@snapchat.com>
Co-authored-by: John Sully <john@csquare.ca>
Co-authored-by: zliang <zliang@snapchat.com>
Co-authored-by: malavan <malavan@eqalpha.com>
Co-authored-by: John Sully <jsully@snapchat.com>
Co-authored-by: jfinity <38383673+jfinity@users.noreply.github.com>
Co-authored-by: benschermel <43507366+benschermel@users.noreply.github.com>
Co-authored-by: guybe7 <guy.benoish@redislabs.com>
Co-authored-by: Karthick Ariyaratnam (A) <k00809413@china.huawei.com>
Co-authored-by: root <paul.chen1@huawei.com>
Co-authored-by: Ilya Shipitsin <chipitsine@gmail.com>
Co-authored-by: Paul Chen <32553156+paulmchen@users.noreply.github.com>
Co-authored-by: Paul Chen <mingchen@Mings-MacBook-Pro.local>
Co-authored-by: Yossi Gottlieb <yossigo@gmail.com>
When a replica paused, it would not apply any commands event the command comes from master, if we feed the non-applied command to replication stream, the replication offset would be wrong, and data would be lost after failover(since replica's `master_repl_offset` grows but command is not applied).
To fix it, here are the changes:
* Don't update replica's replication offset or propagate commands to sub-replicas when it's paused in `commandProcessed`.
* Show `slave_read_repl_offset` in info reply.
* Add an assert to make sure master client should never be blocked unless pause or module (some modules may use block way to do background (parallel) processing and forward original block module command to the replica, it's not a good way but it can work, so the assert excludes module now, but someday in future all modules should rewrite block command to propagate like what `BLPOP` does).
(cherry picked from commit 1b83353dc382959e218191f64d94edb9703552e3)
With an empty src key, we need to deal with two situations:
1. non-STORE: We should return emptyarray.
2. STORE: Try to delete the store key and return 0.
This applies to both GEOSEARCHSTORE (new to v6.2), and
also GEORADIUS STORE (which was broken since forever)
This pr try to fix#9261. i.e. both STORE variants would have behaved
like the non-STORE variants when the source key was missing,
returning an empty array and not deleting the destination key,
instead of returning 0, and deleting the destination key.
Also add more tests for some commands.
- GEORADIUS: wrong type src key, non existing src key, empty search,
store with non existing src key, store with empty search
- GEORADIUSBYMEMBER: wrong type src key, non existing src key,
non existing member, store with non existing src key
- GEOSEARCH: wrong type src key, non existing src key, empty search,
frommember with non existing member
- GEOSEARCHSTORE: wrong type key, non existing src key,
fromlonlat with empty search, frommember with non existing member
Co-authored-by: Oran Agra <oran@redislabs.com>
(cherry picked from commit 86555ae0f7cc45abac7f758d72bf456e90793b46)
- fix possible heap corruption in ziplist and listpack resulting by trying to
allocate more than the maximum size of 4GB.
- prevent ziplist (hash and zset) from reaching size of above 1GB, will be
converted to HT encoding, that's not a useful size.
- prevent listpack (stream) from reaching size of above 1GB.
- XADD will start a new listpack if the new record may cause the previous
listpack to grow over 1GB.
- XADD will respond with an error if a single stream record is over 1GB
- List type (ziplist in quicklist) was truncating strings that were over 4GB,
now it'll respond with an error.
The protocol parsing on 'ldbReplParseCommand' (LUA debugging)
Assumed protocol correctness. This means that if the following
is given:
*1
$100
test
The parser will try to read additional 94 unallocated bytes after
the client buffer.
This commit fixes this issue by validating that there are actually enough
bytes to read. It also limits the amount of data that can be sent by
the debugger client to 1M so the client will not be able to explode
the memory.
This change sets a low limit for multibulk and bulk length in the
protocol for unauthenticated connections, so that they can't easily
cause redis to allocate massive amounts of memory by sending just a few
characters on the network.
The new limits are 10 arguments of 16kb each (instead of 1m of 512mb)
GETBIT, SETBIT may access wrong address because of wrap.
BITCOUNT and BITPOS may return wrapped results.
BITFIELD may access the wrong address but also allocate insufficient memory and segfault (see CVE-2021-32761).
This commit uses `uint64_t` or `long long` instead of `size_t`.
related https://github.com/redis/redis/pull/8096
At 32bit platform:
> setbit bit 4294967295 1
(integer) 0
> config set proto-max-bulk-len 536870913
OK
> append bit "\xFF"
(integer) 536870913
> getbit bit 4294967296
(integer) 0
When the bit index is larger than 4294967295, size_t can't hold bit index. In the past, `proto-max-bulk-len` is limit to 536870912, so there is no problem.
After this commit, bit position is stored in `uint64_t` or `long long`. So when `proto-max-bulk-len > 536870912`, 32bit platforms can still be correct.
For 64bit platform, this problem still exists. The major reason is bit pos 8 times of byte pos. When proto-max-bulk-len is very larger, bit pos may overflow.
But at 64bit platform, we don't have so long string. So this bug may never happen.
Additionally this commit add a test cost `512MB` memory which is tag as `large-memory`. Make freebsd ci and valgrind ci ignore this test.
(cherry picked from commit 71d452876ebf8456afaadd6b3c27988abadd1148)
in case dest key already contains the member, the dest key isn't modified, so the command shouldn't invalidate watch.
(cherry picked from commit 11dc4e59b365d6cd8699604d7d1c1025b6bb6259)
Modules that use background threads with thread safe contexts are likely
to use RM_BlockClient() without a timeout function, because they do not
set up a timeout.
Before this commit, `CLIENT UNBLOCK` would result with a crash as the
`NULL` timeout callback is called. Beyond just crashing, this is also
logically wrong as it may throw the module into an unexpected client
state.
This commits makes `CLIENT UNBLOCK` on such clients behave the same as
any other client that is not in a blocked state and therefore cannot be
unblocked.
(cherry picked from commit aa139e2f02292d668370afde8c91575363c2d611)
SINTERSTORE would have deleted the dest key right away,
even when later on it is bound to fail on an (WRONGTYPE) error.
With this change it first picks up all the input keys, and only later
delete the dest key if one is empty.
Also add more tests for some commands.
Mainly focus on
- `wrong type error`:
expand test case (base on sinter bug) in non-store variant
add tests for store variant (although it exists in non-store variant, i think it would be better to have same tests)
- the dstkey result when we meet `non-exist key (empty set)` in *store
sdiff:
- improve test case about wrong type error (the one we found in sinter, although it is safe in sdiff)
- add test about using non-exist key (treat it like an empty set)
sdiffstore:
- according to sdiff test case, also add some tests about `wrong type error` and `non-exist key`
- the different is that in sdiffstore, we will consider the `dstkey` result
sunion/sunionstore add more tests (same as above)
sinter/sinterstore also same as above ...
(cherry picked from commit b8a5da80c49501773f8778aaf5cbf595cef615e4)
When using RESP3, ZPOPMAX/ZPOPMIN should return nested arrays for consistency
with other commands (e.g. ZRANGE).
We do that only when COUNT argument is present (similarly to how LPOP behaves).
for reasoning see https://github.com/redis/redis/issues/8824#issuecomment-855427955
This is a breaking change only when RESP3 is used, and COUNT argument is present!
(cherry picked from commit 7f342020dcbdf9abe754d6b666efdeded7063870)
- promote the code in DEBUG PROTOCOL to addReplyBigNum
- DEBUG PROTOCOL ATTRIB skips the attribute when client is RESP2
- networking.c addReply for push and attributes generate assertion when
called on a RESP2 client, anything else would produce a broken
protocol that clients can't handle.
(cherry picked from commit 6a5bac309e868deef749c36949723b415de2496f)
due to a copy-paste bug, it used to reply with null response rather than empty array.
this commit includes new tests that are looking at the RESP response directly in
order to be able to tell the difference between them.
Co-authored-by: Oran Agra <oran@redislabs.com>
(cherry picked from commit a418a2d3fc0250c094802d7e8ea64d96eedfda07)
This makes it possible to distinguish between null response and an empty
array (currently the tests infra translates both to an empty string/list)
(cherry picked from commit 7103367ad44b4241e59a709771cb464aa2a86b20)
mistakenly it used to return an empty array rather than 0.
Co-authored-by: Oran Agra <oran@redislabs.com>
(cherry picked from commit 95274f1f8a3ef4cb4033beecfaa99ea1439ed170)
Return a bad score when used with negative count (or count of 1), and non-ziplist encoded zset.
Also add test to validate the return value and cover the issue.
(cherry picked from commit 4bc5a8324d3cb23ed2cc8a9cd19444a893a6d52c)
Previously, passing 0 for newlen would not truncate the string at all.
This adds handling of this case, freeing the old string and creating a new empty string.
Other changes:
- Move `src/modules/testmodule.c` to `tests/modules/basics.c`
- Introduce that basic test into the test suite
- Add tests to cover StringTruncate
- Add `test-modules` build target for the main makefile
- Extend `distclean` build target to clean modules too
(cherry picked from commit 1ccf2ca2f475a161b8ca0c574d4c0e6ef9ecf754)
The decision to stop trimming due to LIMIT in XADD and XTRIM was after the limit was reached.
i.e. the code was deleting **at least** that count of records (from the LIMIT argument's perspective, not the MAXLEN),
instead of **up to** that count of records.
see #9046
(cherry picked from commit eaa7a7bb93c1ac6dbf347e1c29ac719a12a75158)
When test stop 'load handler' by killing the process that generating the load,
some commands that already in the input buffer, still might be processed by the server.
This may cause some instability in tests, that count on that no more commands
processed after we stop the `load handler'
In this commit, new proc 'wait_load_handlers_disconnected' added, to verify that no more
cammands from any 'load handler' prossesed, by checking that the clients who
genreate the load is disconnceted.
Also, replacing check of dbsize with wait_for_ofs_sync before comparing debug digest, as
it would fail in case the last key the workload wrote was an overridden key (not a new one).
Affected tests
Race fix:
- failover command to specific replica works
- Connect multiple replicas at the same time (issue #141), master diskless=$mdl, replica diskless=$sdl
- AOF rewrite during write load: RDB preamble=$rdbpre
Cleanup and speedup:
- Test replication with blocking lists and sorted sets operations
- Test replication with parallel clients writing in different DBs
- Test replication partial resync: $descr (diskless: $mdl, $sdl, reconnect: $reconnect
(cherry picked from commit 32a2584e079a1b3c2d1e6649e38239381a73a459)
There are two issues fixed in this commit:
1. we want to fail the EXEC command in case there is a watched key that's logically
expired but not yet deleted by active expire or lazy expire.
2. we saw that currently cache time is update in every `call()` (including nested calls),
this time is being also being use for the isKeyExpired comparison, we want to update
the cache time only in the first call (execCommand)
Co-authored-by: Oran Agra <oran@redislabs.com>
(cherry picked from commit ac8b1df8850cc80fbf9ce8c2fbde0c1d3a1b4e91)
The `Tracking gets notification of expired keys` test in tracking.tcl
used to hung in valgrind CI quite a lot.
It turns out the reason is that with valgrind and a busy machine, the
server cron active expire cycle could easily run in the same event loop
as the command that created `mykey`, so that when they key got expired,
there were two change events to broadcast, one that set the key and one
that expired it, but since we used raxTryInsert, the client that was
associated with the "last" change was the one that created the key, so
the NOLOOP filtered that event.
This commit adds a test that reproduces the problem by using lazy expire
in a multi-exec which makes sure the key expires in the same event loop
as the one that added it.
(cherry picked from commit 9b564b525d8ce88295ec14ffdc3bede7e5f5c33e)
