2753 Commits

Author SHA1 Message Date
antirez
e46942e4b0 Sentinel: handle Hello messages received via slaves correctly.
Even when messages are received via the slave, we should perform
operations (like adding a new Sentinel) in the context of the master.
2013-11-11 17:12:27 +01:00
antirez
3d7000f281 Sentinel: remove code not useful in the new design. 2013-11-11 12:06:11 +01:00
antirez
49170f6c12 Sentinel: epoch introduced.
Sentinel state now includes the idea of current epoch and config epoch.
In the Hello message, that is now published both on masters and slaves,
a Sentinel no longer just advertises itself but also broadcasts its
current view of the configuration: the master name / ip / port and its
current epoch.

Sentinels receiving such information switch to the new master if the
configuration epoch received is newer and the ip / port of the master
are indeed different compared to the previos ones.
2013-11-11 11:05:58 +01:00
antirez
c46f655c90 Log to what master a slave is going to connect to. 2013-11-11 09:25:36 +01:00
antirez
e159239f9c Cluster: removed not needed newline at end of redisLog() msg. 2013-11-08 17:28:02 +01:00
antirez
a67935e5e3 Cluster: send a single UPDATE packet for now. 2013-11-08 17:25:49 +01:00
antirez
a146482c83 Cluster: replace hardcoded 4096 for bus msg len with sizeof(). 2013-11-08 17:19:19 +01:00
antirez
36db83ac50 Cluster: slots update refactored + UPDATE msg processing.
Now there is a function that handles the update of the local slot
configuration every time we have some new info about a node and its set
of served slots and configEpoch.

Moreoever the UPDATE packets are now processed when received (it was a
work in progress in the previous commit).
2013-11-08 17:02:10 +01:00
antirez
a19147c2fa Cluster: UPDATE msg data structure and sending function. 2013-11-08 16:26:50 +01:00
antirez
4666966c9f Cluster: refactoring of slots update code and more.
The commit also introduces detection of nodes publishing not updated
configuration. More work in progress to send an UPDATE packet to inform
of the config change.
2013-11-08 10:32:16 +01:00
antirez
8c2127f9c9 Fix broken rdbWriteRaw() return value check in rdb.c.
Thanks to @PhoneLi for reporting.
2013-11-07 23:53:18 +01:00
antirez
605ba6ad38 redis-trib: fixed slot allocation when --replicas is used. 2013-11-07 16:12:06 +01:00
antirez
b64dbf7387 Sentinel: sentinelSendSlaveOf() was missing a var and the prototype. 2013-11-06 11:23:53 +01:00
antirez
3be2b18d9f Sentinel: increment pending_commands counter in two more places.
AUTH and SCRIPT KILL were sent without incrementing the pending commands
counter. Clearly this needs some kind of wrapper doing it for the caller
in order to be less bug prone.
2013-11-06 11:21:44 +01:00
antirez
1071abc4a4 Sentinel: always send CONFIG REWRITE when changing instance role.
This change makes Sentinel less fragile about a number of failure modes.

This commit also fixes a different bug as a side effect, SLAVEOF command
was sent multiple times without incrementing the pending commands count.
2013-11-06 11:13:27 +01:00
antirez
1a1eb8bc8d SCAN code refactored to parse cursor first.
The previous implementation of SCAN parsed the cursor in the generic
function implementing SCAN, SSCAN, HSCAN and ZSCAN.

The actual higher-level command implementation only checked for empty
keys and return ASAP in that case. The result was that inverting the
arguments of, for instance, SSCAN for example and write:

    SSCAN 0 key

Instead of

    SSCAN key 0

Resulted into no error, since 0 is a non-existing key name very likely.
Just the iterator returned no elements at all.

In order to fix this issue the code was refactored to extract the
function to parse the cursor and return the error. Every higher level
command implementation now parses the cursor and later checks if the key
exist or not.
2013-11-05 15:47:50 +01:00
antirez
f31cf249b9 SCAN: when iterating ziplists or intsets always return cursor of 0.
The previous implementation assumed that the first call always happens
with cursor set to 0, this may not be the case, and we want to return 0
anyway otherwise the (broken) client code will loop forever.
2013-11-05 15:32:25 +01:00
antirez
b97cff8b63 Use strtoul() instead of sscanf() in SCAN implementation. 2013-11-05 15:30:21 +01:00
antirez
56f53f7b3c HSCAN/ZSCAN: skip value when matching.
This fixes issue #1360 and #1362.
2013-11-05 12:16:29 +01:00
antirez
f6738923a6 Cluster: initialize senderConfigEpoch and senderCurrentEpoch for warnings suppression. 2013-11-05 12:01:07 +01:00
antirez
522549729f Pass int64_t to intsetGet() instead of long long. 2013-11-05 11:57:30 +01:00
antirez
865d3b0f33 removed not used vars in dictScan(). 2013-11-05 11:56:11 +01:00
antirez
e1a469d3b0 Initial support for --replicas in redis-trib. 2013-11-05 11:24:24 +01:00
antirez
653347d253 Inverted variable boolean value and name after scanGenericCommand() refactoring. 2013-10-31 10:35:56 +01:00
antirez
34e471e748 scanGenericCommand() refactoring and handling of integer encoded elements.
This commit fixes issue #1354.
2013-10-31 10:32:39 +01:00
antirez
af411f4c92 redis-benchmark: update help for new __rand_int__ form. 2013-10-28 18:13:42 +01:00
antirez
e23fa0ec99 Aesthetic fix (missing space) into HSCAN and ZSCAN implementations.
Thanks to @badboy for reporting.
2013-10-28 13:20:11 +01:00
antirez
b2618c6cdb ZSCAN implemented. 2013-10-28 11:36:42 +01:00
antirez
6618167a9f HSCAN implemented. 2013-10-28 11:35:26 +01:00
antirez
e96ffac563 SSCAN implemented. 2013-10-28 11:17:32 +01:00
antirez
99efa37a6b dictScan(): empty hash table requires special handling. 2013-10-28 11:17:18 +01:00
antirez
e74c235567 SCAN is a random command and does not require output sorting.
Sorting the output helps when we want to turn a non-deterministic into a
deterministic command, in that case this is not possible.
2013-10-28 11:13:43 +01:00
antirez
0685c1ca13 SCAN: refactored into scanGenericCommand.
The new implementation is capable of iterating the keyspace but also
sets, hashes, and sorted sets, and can be used to implement SSCAN, ZSCAN
and HSCAN.
2013-10-28 11:11:34 +01:00
antirez
f86c07df30 Fixed typos in dictScan() comment. 2013-10-25 17:05:55 +02:00
antirez
817e6766aa dictScan() algorithm documented. 2013-10-25 17:01:30 +02:00
antirez
eda292a7fa SCAN: stay inside 80 cols. 2013-10-25 12:01:49 +02:00
antirez
fe0ffe6a1c Revert "Fixed typo in SCAN comment. iff -> if."
Probably here Pieter means "if and only if".

This reverts commit 0a970b93e911c0ece1fa2551ade71e122d8c01e1.
2013-10-25 12:00:13 +02:00
antirez
7bd87839c1 SCAN: simplify keys list cleanup using listSetFreeMethod(). 2013-10-25 11:58:03 +02:00
antirez
6f69128751 SCAN: improve variable names for readability. 2013-10-25 11:54:45 +02:00
antirez
6bff0f3cb6 SCAN: remove additional newlines to conform to Redis code base. 2013-10-25 11:51:08 +02:00
antirez
32b555c5fb SCAN: remove useless assertion, already enforced by command table. 2013-10-25 11:49:08 +02:00
antirez
6874fcc216 SCAN: use define REDIS_LONGSTR_SIZE instead of fixed len. 2013-10-25 11:48:18 +02:00
antirez
0a970b93e9 Fixed typo in SCAN comment. iff -> if. 2013-10-25 11:46:02 +02:00
antirez
5227a1f66f SCAN option name changed: pattern -> match. 2013-10-25 11:45:32 +02:00
Pieter Noordhuis
f18269d1ef Fix error in scan algorithm
The irrelevant bits shouldn't be masked to 1. This can result in slots being
skipped when the hash table is resized between calls to the iterator.
2013-10-25 10:50:03 +02:00
Pieter Noordhuis
25ae316f65 SCAN requires at least 1 argument 2013-10-25 10:49:56 +02:00
Pieter Noordhuis
956c0ed927 Add SCAN command 2013-10-25 10:49:48 +02:00
antirez
0dbe09bfec Cluster: rough support for sub-command options in redis-trib. 2013-10-11 17:33:19 +02:00
antirez
e45d9420e0 Cluster: there is a lower limit for the handshake timeout. 2013-10-11 10:34:32 +02:00
antirez
39c90945e0 Cluster: data_age conversion to milliseconds fixed. 2013-10-09 16:36:06 +02:00