gvsafronov/futriix
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
9,953 Commits 1 Branch 0 Tags
Commit Graph

7142 Commits

Author SHA1 Message Date
John Sully
a060bc7942 New expire datastructure and algorithm. Allows us to expire in sublinear time 
Former-commit-id: 3880d2616c882e19169180dc10268564347b0279
 2019-07-23 18:30:10 -04:00
Salvatore Sanfilippo
ece91a4bc7 Merge pull request #6265 from madolson/dev-unstable-remove-create-object 
Removed unnecessary creation of Redis objects
 2019-07-23 17:05:10 +02:00
antirez
1d54931685 Client side caching: show tracking slots usage in INFO. 2019-07-23 11:02:14 +02:00
antirez
79a73bb95a Client side caching: implement full slot limit function. 2019-07-23 10:57:22 +02:00
zhaozhao.zz
408ca31d16 Client side caching: do not reclaim tracking table if it's empty 2019-07-23 15:25:00 +08:00
Madelyn Olson
4520e87800 Hide HELLO and AUTH from slowlog and monitor 2019-07-22 22:53:15 -07:00
Madelyn Olson
dedbc9d4d2 Removed unecessary creation of Redis objects 2019-07-22 22:52:16 -07:00
Oran Agra
ff0780e8e6 Implement module api for aux data in rdb 
Other changes:
* fix memory leak in error handling of rdb loading of type OBJ_MODULE
 2019-07-22 21:15:33 +03:00
antirez
f2f3d91e78 Client side caching: split invalidation into key / slot. 2019-07-22 18:59:53 +02:00
antirez
023b67e80f Client side caching: redis-cli ability to enable tracking. 
This is extremely useful in order to simulate an high load of requests
about different keys, and force Redis to track a lot of informations
about several clients, to simulate real world workloads.
 2019-07-22 18:45:47 +02:00
antirez
ac05029790 Move signalFlushedDb() into a better place. 
Now that the call also invalidates client side caching slots, it is
important that after an internal flush operation we both send the
notifications to the clients and, at the same time, are able to reclaim
the memory of the tracking table. This may even fix a few edge cases
related to MULTI/EXEC + WATCH during resync, not sure, but in general
looks more correct.
 2019-07-22 12:37:47 +02:00
antirez
06de8790dc Client side caching: reclaim the tracking table on FLUSHALL. 2019-07-22 12:33:49 +02:00
antirez
df2e7cf4fc Client side caching: call the invalidation functions always. 
Otherwise what happens is that the tracking table will never get garbage
collected if there are no longer clients with tracking enabled.
Now the invalidation function immediately checks if there is any table
allocated, otherwise it returns ASAP, so the overhead when the feature
is not used should be near zero.
 2019-07-22 12:29:54 +02:00
antirez
ab49072362 Client side caching: don't hash the key if not needed. 2019-07-22 12:10:51 +02:00
antirez
bcf3cf8832 Clinet side caching: take count of used caching slots. 2019-07-22 11:47:44 +02:00
chendianqiang
4b89b52355 make memory usage consistent of robj with OBJ_ENCODING_INT 2019-07-22 17:45:30 +08:00
Oran Agra
c80ad2f4b5 Allow modules to handle RDB loading errors. 
This is especially needed in diskless loading, were a short read could have
caused redis to exit. now the module can handle the error and return to the
caller gracefully.

this fixes #5326
 2019-07-21 18:19:32 +03:00
John Sully
174016dee7 Modules must have execute permissions to load 2019-07-19 15:28:31 -04:00
John Sully
f094402c33 Any +x bit is acceptable 
Former-commit-id: 156e596f9c7a922bc3361652b74b78bbeab0f2dc
 2019-07-19 15:26:17 -04:00
antirez
92455bb036 RDB: fix MODULE_AUX loading by continuing to next opcode. 
Thanks to @JohnSully for noticing this problem.
 2019-07-19 11:12:46 +02:00
John Sully
8d526d9354 Any +x bit is acceptable 
Former-commit-id: fc58516cca72fc9db97bc4c388f9fa692d115df4
 2019-07-19 01:42:05 -04:00
John Sully
38205b5526 Modules must have execute permissions to load 2019-07-19 01:37:34 -04:00
John Sully
5404d6f6bb Modules must have execute permissions to load 
Former-commit-id: a4efcd35af52227a22daf7f882e8e14db3f8bf57
 2019-07-19 01:31:10 -04:00
antirez
ebc50797a0 RDB: make sure to abort on LZF encoding error. 
Former-commit-id: 27fe1658a2019bcd5d880e844bac21ccef8303f2
 2019-07-19 01:03:16 -04:00
antirez
867070eb99 RDB: handle encoding errors with rdbExitReportCorruptRDB(). 
Without such change, the diskless replicas, when loading RDB files from
the socket will not abort when a broken RDB file gets loaded. This is
potentially unsafe, because right now Redis is not able to guarantee
that encoding errors are safe from the POV of memory corruptions (for
instance the LZF library may not be safe against untrusted data?) so
better to abort when the RDB file we are going to load is corrupted.

Instead I/O errors are still returned to the caller without aborting,
so that in case of short read the diskless replica can try again.


Former-commit-id: 47feb2719ca7fd04e7e108ec1af0f777e536bf8a
 2019-07-19 01:02:49 -04:00
John Sully
1c8c4a5db2 Fix bad merge in SCAN KEYS command 
Former-commit-id: c21af6b351328ffbdb1d1e2a7eed44f8f929f8b2
 2019-07-19 00:43:23 -04:00
John Sully
c178b67ded Merge commit '9c1932beeb6e307c2fbeadcfff8954e517189ae8' into unstable 
Former-commit-id: 821c12c482d20b15dfb5a6eeab52e167997627d8
 2019-07-19 00:32:43 -04:00
John Sully
33acb40633 Merge commit '985e5b2c608eade2a60e50a6a177f13381c9c8d8' into unstable 
Former-commit-id: 4602f8c391409e9dd59f1fbee6a5ef011b219ca1
 2019-07-19 00:01:56 -04:00
John Sully
6e461c3cc3 Fix compile errors from merges 
Former-commit-id: 27a927fe0011536c6539d7c2a79ccfdaf78cee22
 2019-07-18 23:35:51 -04:00
Angus Pearson
12c8823c93 Add char* typeNameCanonicalize(robj*) to remove duplicate code between SCAN and TYPE commands, 
and to keep OBJ_* enum to string canonicalization in one place.


Former-commit-id: 3cdc6e8d846e88cf4e250b2643662bde2a9317c5
 2019-07-18 23:31:31 -04:00
Madelyn Olson
a25866ad35 Fixed some spelling issues in ACL codepath including user facing error 
Former-commit-id: 50ad880ad55e7761fe2598e09be43947e88740fe
 2019-07-18 23:28:36 -04:00
John Sully
11b12219eb Merge commit '214555b02ea448c444fb6a32ecfc5f312fc052b8' into unstable 
Former-commit-id: 15bed4bbcc962ec1e8ad5f57653534b721df9a70
 2019-07-18 23:26:46 -04:00
John Sully
13fa9bb2be Merge commit '4bbaf621a12f6ec22ec7dad6a2282b7908660497' into unstable 
Former-commit-id: 10986ae3a801c13013ec74880dd1f7969c284d01
 2019-07-18 23:22:36 -04:00
John Sully
6f3eda55d9 Merge commit '9eea57cc311e62a49358e09af7baebce9da9053f' into unstable 
Former-commit-id: ee460301fd160cf3fc82e52cf47d4a5f7503d781
 2019-07-18 23:20:46 -04:00
John Sully
79b33481ae Merge commit '3f4f7aff1aa6abf8cad1726941d4c33b3bfbc9b8' into unstable 
Former-commit-id: 8ebe2fce3da506639acb37fedf1a2d3e47669572
 2019-07-18 23:18:42 -04:00
antirez
b2d03d4221 Make comment in getClientOutputBufferMemoryUsage() describing the present. 
Former-commit-id: 35acae360a4c3c67370b03b4835c51b08194ca28
 2019-07-18 23:12:39 -04:00
John Sully
40eefe7da7 Release version 5 
Former-commit-id: 25e18a7de91d4af651f8f22674846cb0238dd5d3
 2019-07-18 20:50:40 -04:00
John Sully
cfee6376d2 Active Replicas should always use their own slaveseldb variable when sending/saving rdbs - never those of their masters. This is because the Active Replica is also simultaneously a master 
Former-commit-id: bb090b6d20bfc3d5d3ca07270f64bf15fad42681
 2019-07-18 20:50:20 -04:00
John Sully
5cb7ec841a Add support for sanitizers and fix issues they have found 
Former-commit-id: 58971a430dd3e4e8c39c7b5288211bc2ba7699ed
 2019-07-18 18:39:42 -04:00
John Sully
bd3cc0bb95 Resolve race on neterr variable 
Former-commit-id: 1512274f4c51dc4e8f4e3d11df2f35e051c0c079
 2019-07-18 17:11:28 -04:00
John Sully
857f778b38 Make object refcounts atomic, technically it doesn't need to be currently but the semantics are complicated and error prone. This is much safer with little performance impact 
Former-commit-id: e310d33c121550f69b1c06d101db0c3f944a7fb0
 2019-07-18 15:49:12 -04:00
John Sully
ce595a9287 Suppress initialized data sanitizer warnings 
Former-commit-id: ad32ba0343a241c46206ada273b0c0125e9e3101
 2019-07-18 15:49:03 -04:00
antirez
b8d5518787 RDB: make sure to abort on LZF encoding error. 2019-07-18 18:59:38 +02:00
antirez
a345fef285 RDB: handle encoding errors with rdbExitReportCorruptRDB(). 
Without such change, the diskless replicas, when loading RDB files from
the socket will not abort when a broken RDB file gets loaded. This is
potentially unsafe, because right now Redis is not able to guarantee
that encoding errors are safe from the POV of memory corruptions (for
instance the LZF library may not be safe against untrusted data?) so
better to abort when the RDB file we are going to load is corrupted.

Instead I/O errors are still returned to the caller without aborting,
so that in case of short read the diskless replica can try again.
 2019-07-18 18:51:45 +02:00
Salvatore Sanfilippo
344f46380f Merge pull request #6252 from soloestoy/tracking-flushdb 
Tracking flushdb
 2019-07-18 16:41:37 +02:00
antirez
db69845b48 RDB: update rdbLoadRio comment about EOF condition. 2019-07-18 12:37:55 +02:00
zhaozhao.zz
996dcae44f Client side caching: filter clients untracking 2019-07-18 16:46:19 +08:00
John Sully
48e44624c3 Another RESP3 regression 
Former-commit-id: 4e30a5b4e53a4d1580f8d7cca8a8eac11e36ce4d
 2019-07-17 22:57:52 -04:00
antirez
72b7b904c1 RDB: try to make error handling code more readable. 2019-07-17 17:30:02 +02:00
antirez
94186a607e Rio: when in error condition avoid doing the operation. 2019-07-17 16:46:22 +02:00