Merge "Make statsd sampling rates configurable"

diff --git a/includes/DefaultSettings.php b/includes/DefaultSettings.php
index 2b55f1b..a990525 100644 (file)
--- a/includes/DefaultSettings.php
+++ b/includes/DefaultSettings.php
@@ -6162,6 +6162,14 @@ $wgStatsdServer = false;
  */
 $wgStatsdMetricPrefix = 'MediaWiki';
 
+/**
+ * Sampling rate for statsd metrics as an associative array of patterns and rates.
+ * Patterns are Unix shell patterns (e.g. 'MediaWiki.api.*').
+ * Rates are sampling probabilities (e.g. 0.1 means 1 in 10 events are sampled).
+ * @since 1.28
+ */
+$wgStatsdSamplingRates = [];
+
 /**
  * InfoAction retrieves a list of transclusion links (both to and from).
  * This number puts a limit on that query in the case of highly transcluded

diff --git a/includes/GlobalFunctions.php b/includes/GlobalFunctions.php
index 0d66908..b421f96 100644 (file)
--- a/includes/GlobalFunctions.php
+++ b/includes/GlobalFunctions.php
@@ -1195,6 +1195,7 @@ function wfLogProfilingData() {
                        $statsdPort = isset( $statsdServer[1] ) ? $statsdServer[1] : 8125;
                        $statsdSender = new SocketSender( $statsdHost, $statsdPort );
                        $statsdClient = new SamplingStatsdClient( $statsdSender, true, false );
+                       $statsdClient->setSamplingRates( $config->get( 'StatsdSamplingRates' ) );
                        $statsdClient->send( $context->getStats()->getBuffer() );
                } catch ( Exception $ex ) {
                        MWExceptionHandler::logException( $ex );

diff --git a/includes/libs/SamplingStatsdClient.php b/includes/libs/SamplingStatsdClient.php
index 2e780c9..dd1976c 100644 (file)
--- a/includes/libs/SamplingStatsdClient.php
+++ b/includes/libs/SamplingStatsdClient.php
@@ -30,6 +30,19 @@ use Liuggio\StatsdClient\Entity\StatsdDataInterface;
  * @since 1.26
  */
 class SamplingStatsdClient extends StatsdClient {
+       protected $samplingRates = [];
+
+       /**
+        * Sampling rates as an associative array of patterns and rates.
+        * Patterns are Unix shell patterns (e.g. 'MediaWiki.api.*').
+        * Rates are sampling probabilities (e.g. 0.1 means 1 in 10 events are sampled).
+        * @param array $samplingRates
+        * @since 1.28
+        */
+       public function setSamplingRates( array $samplingRates ) {
+               $this->samplingRates = $samplingRates;
+       }
+
        /**
         * Sets sampling rate for all items in $data.
         * The sample rate specified in a StatsdData entity overrides the sample rate specified here.
@@ -37,11 +50,18 @@ class SamplingStatsdClient extends StatsdClient {
         * {@inheritDoc}
         */
        public function appendSampleRate( $data, $sampleRate = 1 ) {
-               if ( $sampleRate < 1 ) {
-                       array_walk( $data, function( $item ) use ( $sampleRate ) {
+               $samplingRates = $this->samplingRates;
+               if ( !$samplingRates && $sampleRate !== 1 ) {
+                       $samplingRates = [ '*' => $sampleRate ];
+               }
+               if ( $samplingRates ) {
+                       array_walk( $data, function( $item ) use ( $samplingRates ) {
                                /** @var $item StatsdData */
-                               if ( $item->getSampleRate() === 1 ) {
-                                       $item->setSampleRate( $sampleRate );
+                               foreach ( $samplingRates as $pattern => $rate ) {
+                                       if ( fnmatch( $pattern, $item->getKey(), FNM_NOESCAPE ) ) {
+                                               $item->setSampleRate( $item->getSampleRate() * $rate );
+                                               break;
+                                       }
                                }
                        } );
                }
@@ -74,9 +94,7 @@ class SamplingStatsdClient extends StatsdClient {
                }
 
                // add sampling
-               if ( $sampleRate < 1 ) {
-                       $data = $this->appendSampleRate( $data, $sampleRate );
-               }
+               $data = $this->appendSampleRate( $data, $sampleRate );
                $data = $this->sampleData( $data );
 
                $data = array_map( 'strval', $data );

diff --git a/tests/phpunit/includes/libs/SamplingStatsdClientTest.php b/tests/phpunit/includes/libs/SamplingStatsdClientTest.php
index 1ebe551..9a48930 100644 (file)
--- a/tests/phpunit/includes/libs/SamplingStatsdClientTest.php
+++ b/tests/phpunit/includes/libs/SamplingStatsdClientTest.php
@@ -32,12 +32,35 @@ class SamplingStatsdClientTest extends PHPUnit_Framework_TestCase {
 
                return [
                        // $data, $sampleRate, $seed, $expectWrite
-                       [ $unsampled, 1, 0 /*0.44*/, $unsampled ],
-                       [ $sampled, 1, 0 /*0.44*/, null ],
-                       [ $sampled, 1, 4 /*0.03*/, $sampled ],
-                       [ $unsampled, 0.1, 4 /*0.03*/, $sampled ],
-                       [ $sampled, 0.5, 0 /*0.44*/, null ],
-                       [ $sampled, 0.5, 4 /*0.03*/, $sampled ],
+                       [ $unsampled, 1, 0 /*0.44*/, true ],
+                       [ $sampled, 1, 0 /*0.44*/, false ],
+                       [ $sampled, 1, 4 /*0.03*/, true ],
+                       [ $unsampled, 0.1, 0 /*0.44*/, false ],
+                       [ $sampled, 0.5, 0 /*0.44*/, false ],
+                       [ $sampled, 0.5, 4 /*0.03*/, false ],
                ];
        }
+
+       public function testSetSamplingRates() {
+               $matching = new StatsdData();
+               $matching->setKey( 'foo.bar' );
+               $matching->setValue( 1 );
+
+               $nonMatching = new StatsdData();
+               $nonMatching->setKey( 'oof.bar' );
+               $nonMatching->setValue( 1 );
+
+               $sender = $this->getMock( 'Liuggio\StatsdClient\Sender\SenderInterface' );
+               $sender->expects( $this->any() )->method( 'open' )->will( $this->returnValue( true ) );
+               $sender->expects( $this->once() )->method( 'write' )->with( $this->anything(),
+                       $this->equalTo( $nonMatching ) );
+
+               $client = new SamplingStatsdClient( $sender );
+               $client->setSamplingRates( [ 'foo.*' => 0.2 ] );
+
+               mt_srand( 0 ); // next random is 0.44
+               $client->send( $matching );
+               mt_srand( 0 );
+               $client->send( $nonMatching );
+       }
 }