gvsafronov/futriix
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
4,452 Commits 1 Branch 0 Tags
Commit Graph

4452 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
antirez
49903897ef LATENCY HISTORY returns empty array if no data is available. 
Previously we returned an error.
 2014-07-10 16:20:40 +02:00
antirez
186b5b0f66 Test: enable latency monitor in the default config. 
This way as a side effect of running the test we also stress the latency
monitor data collection.
 2014-07-10 15:04:34 +02:00
antirez
383536119d Test: AOF rewrite during write load. 2014-07-10 11:25:12 +02:00
antirez
a1f3072ab5 Fixed a few missing newline in createLatencyReport(). 2014-07-09 16:45:40 +02:00
antirez
cdc44eaf1a createLatencyReport(): compile before commit avoids commits. 2014-07-08 17:17:08 +02:00
antirez
8dfa279c19 createLatencyReport(), fix mount option name. 2014-07-08 17:12:21 +02:00
antirez
5102269c5a createLatencyReport(): initialize all advices to zero. 2014-07-08 17:11:56 +02:00
antirez
a519c133a6 LATENCY DOCTOR first implementation complete. 2014-07-08 17:05:56 +02:00
antirez
d1e9441165 Fix analyzeLatencyForEvent() MAD computation. 2014-07-08 12:56:45 +02:00
antirez
b65c20e3ea LATENCY DOCTOR: initial draft and events summary output. 2014-07-08 11:31:46 +02:00
antirez
608c0051d6 Latency: low level time series analysis implemented. 2014-07-07 15:00:01 +02:00
antirez
84018db59c LATENCY SAMPLES renamed LATENCY HISTORY. 2014-07-07 12:35:24 +02:00
antirez
9269ddff7c LATENCY RESET implemented. 2014-07-07 12:34:54 +02:00
antirez
378e020b71 Better "final read from parent" algorithm in rewriteAppendOnlyFile*(. 
We now wait up to 1 second for diff data to come from the parent,
however we use poll(2) to wait for more data, and use a counter of
contiguous failures to get data for N times (set to 20 experimentally
after different tests) as an early stop condition to avoid wasting 1
second when the write traffic is too low.
 2014-07-05 15:42:24 +02:00
antirez
3d745b4919 Log AOF diff sizes in MBs instead of bytes. 2014-07-05 12:37:44 +02:00
antirez
8a89fd0b46 aofChildWriteDiffData() better handling of free blocks. 2014-07-05 01:11:28 +02:00
antirez
26216ed305 redis-benchmark: abort when all clients are disconnected. 2014-07-04 17:52:18 +02:00
antirez
b8f4f661f1 Fine tuning of aofReadDiffFromParent() calls trigger. 2014-07-04 17:51:03 +02:00
antirez
9d302ec21e Use amount of I/O not ops to trigger aofReadDiffFromParent(). 2014-07-04 16:03:46 +02:00
antirez
1937f8a8b0 aofChildWriteDiffData(): write as much as possible. 2014-07-04 15:54:20 +02:00
antirez
f4e54913d6 Handle write failure in AOF parent -> child ACK. 2014-07-04 15:37:49 +02:00
antirez
cfa0cad4af Use a timeout when reading parent ack from AOF child. 2014-07-04 15:35:01 +02:00
antirez
3f77583764 Send AOF diffs from parent to child to improve latency. 2014-07-04 15:27:23 +02:00
antirez
1e8520e84a Use fsync instead of aof_fsync in final AOF sync. 
This happens in the child process so we don't care about latency:
better to sync metadata as well.
 2014-07-04 11:51:30 +02:00
antirez
3a9acbdb5d Fixed conditional for aof-write-pending-fsync latency event selection. 2014-07-02 17:42:29 +02:00
antirez
9bdfff137f Dependencies updated. 2014-07-02 17:04:14 +02:00
antirez
db90e562e7 Cast void* to char* to avoid waring in latencyCommand(). 2014-07-02 16:56:08 +02:00
antirez
0954e4ede9 Properly initialize min/max in latency.c. 2014-07-02 16:55:34 +02:00
antirez
4a89da385c latencyStartMonitor() modified to avoid warnings. 2014-07-02 16:53:44 +02:00
antirez
2dfe2d9493 Latency monitor: specialize delayed aof writes events. 2014-07-02 16:45:45 +02:00
antirez
d13384e7d3 LATENCY GRAPH: filling under the curve is more readable. 2014-07-02 16:37:53 +02:00
antirez
0039e2e996 LATENCY GRAPH implemented. 2014-07-02 16:31:22 +02:00
antirez
1760e6ede1 latencyTimeSeries structure max field type fixed. 2014-07-02 16:14:28 +02:00
antirez
b55b72c3e1 Free labels in freeSparklineSequence(). 2014-07-02 12:49:14 +02:00
antirez
2127c8e8fa LATENCY LATEST: add the max field. 2014-07-02 12:40:38 +02:00
antirez
71c60780e1 Latency monitor trheshold value is now configurable. 
This commit adds both support for redis.conf and CONFIG SET/GET.
 2014-07-02 12:28:17 +02:00
antirez
fa7679c047 ASCII sparklines generation API. 2014-07-02 10:13:53 +02:00
antirez
acb5c58fbb License added to latency.h. 2014-07-02 10:06:58 +02:00
antirez
084f154c94 Latency monitor turned off by default. 
It is not a good idea to bloat the code with gettimeofday() calls if the
instance is working well, and turning monitoring on at runtime is a
joke.
 2014-07-01 17:23:59 +02:00
antirez
51116b4638 Latency monitor: more hooks around the code. 2014-07-01 17:19:08 +02:00
antirez
f765e5a698 Latency monitor: don't add new samples in the same second. 
Instead we update the old sample with the new latency if it is greater.
 2014-07-01 17:12:09 +02:00
antirez
224b5add47 LATENCY LATEST implemented. 2014-07-01 16:17:33 +02:00
antirez
69ba6924c9 Latency monitor: command duration is in useconds. Convert. 2014-07-01 16:09:02 +02:00
antirez
4ef47b48b0 LATENCY SAMPLES implemented. 2014-07-01 16:07:13 +02:00
antirez
47f819f87d Latency monitor: collect slow commands. 
We introduce the distinction between slow and fast commands since those
are two different sources of latency. An O(1) or O(log N) command without
side effects (can't trigger deletion of large objects as a side effect of
its execution) if delayed is a symptom of inherent latency of the system.

A non-fast command (commands that may run large O(N) computations) if
delayed may just mean that the user is executing slow operations.

The advices LATENCY should provide in this two different cases are
different, so we log the two classes of commands in a separated way.
 2014-07-01 11:47:08 +02:00
antirez
eabdfb3e30 Latency monitor: basic samples collection. 2014-07-01 11:30:15 +02:00
antirez
3513d5a237 Fix Solaris compilation due to ctime_r() call. 
Introduced in Redis 2.8.10 because of a change in Sentinel.
This closes issue #1837.
 2014-06-30 16:29:12 +02:00
antirez
e2078f8001 Test: find_available_port: check that cluster port is free as well. 
The function will only return ports that have also port+10000 free, so
that Redis Cluster instances can be executed at the returned port.
 2014-06-30 12:08:24 +02:00
antirez
91c02dff33 Test: fix instances.tcl restart_instance abort error. 2014-06-30 12:06:27 +02:00
antirez
ff58279939 DEBUG CMDKEYS moved to COMMAND GETKEYS. 2014-06-27 12:22:15 +02:00