mirror of
https://github.com/osm-search/Nominatim.git
synced 2026-02-15 19:07:58 +00:00
prune list of blocked items if too large, different costs for different page types
This commit is contained in:
@@ -4,6 +4,7 @@
|
||||
|
||||
require_once(CONST_BasePath.'/settings/settings.php');
|
||||
require_once(CONST_BasePath.'/lib/lib.php');
|
||||
require_once(CONST_BasePath.'/lib/leakybucket.php');
|
||||
require_once(CONST_BasePath.'/lib/db.php');
|
||||
|
||||
if (get_magic_quotes_gpc())
|
||||
|
||||
137
lib/leakybucket.php
Normal file
137
lib/leakybucket.php
Normal file
@@ -0,0 +1,137 @@
|
||||
<?php
|
||||
|
||||
function getBucketMemcache()
|
||||
{
|
||||
static $m;
|
||||
|
||||
if (!CONST_ConnectionBucket_MemcacheServerAddress) return null;
|
||||
if (!isset($m))
|
||||
{
|
||||
$m = new Memcached();
|
||||
$m->addServer(CONST_ConnectionBucket_MemcacheServerAddress, CONST_ConnectionBucket_MemcacheServerPort);
|
||||
}
|
||||
return $m;
|
||||
}
|
||||
|
||||
function doBucket($asKey, $iRequestCost, $iLeakPerSecond, $iThreshold)
|
||||
{
|
||||
$m = getBucketMemcache();
|
||||
if (!$m) return 0;
|
||||
|
||||
$iMaxVal = 0;
|
||||
$t = time();
|
||||
|
||||
foreach($asKey as $sKey)
|
||||
{
|
||||
$aCurrentBlock = $m->get($sKey);
|
||||
if (!$aCurrentBlock)
|
||||
{
|
||||
$aCurrentBlock = array($iRequestCost, $t);
|
||||
}
|
||||
else
|
||||
{
|
||||
// add RequestCost
|
||||
// remove leak * the time since the last request
|
||||
$aCurrentBlock[0] += $iRequestCost - ($t - $aCurrentBlock[1])*$iLeakPerSecond;
|
||||
$aCurrentBlock[1] = $t;
|
||||
}
|
||||
|
||||
if ($aCurrentBlock[0] <= 0)
|
||||
{
|
||||
$m->delete($sKey);
|
||||
}
|
||||
else
|
||||
{
|
||||
// If we have hit the threshold stop and record this to the block list
|
||||
if ($aCurrentBlock[0] >= $iThreshold)
|
||||
{
|
||||
$aCurrentBlock[0] = $iThreshold;
|
||||
|
||||
// Make up to 10 attempts to record this to memcache (with locking to prevent conflicts)
|
||||
$i = 10;
|
||||
for($i = 0; $i < 10; $i++)
|
||||
{
|
||||
$aBlockedList = $m->get('blockedList', null, $hCasToken);
|
||||
if (!$aBlockedList)
|
||||
{
|
||||
$aBlockedList = array();
|
||||
$m->add('blockedList', $aBlockedList);
|
||||
$aBlockedList = $m->get('blockedList', null, $hCasToken);
|
||||
}
|
||||
if (!isset($aBlockedList[$sKey]))
|
||||
{
|
||||
$aBlockedList[$sKey] = array(1, $t);
|
||||
}
|
||||
else
|
||||
{
|
||||
$aBlockedList[$sKey][0]++;
|
||||
$aBlockedList[$sKey][1] = $t;
|
||||
}
|
||||
if (sizeof($aBlockedList) > CONST_ConnectionBucket_MaxBlockList)
|
||||
{
|
||||
uasort($aBlockedList, 'byValue1');
|
||||
$aBlockedList = array_slice($aBlockedList, 0, CONST_ConnectionBucket_MaxBlockList);
|
||||
}
|
||||
$x = $m->cas($hCasToken, 'blockedList', $aBlockedList);
|
||||
if ($x) break;
|
||||
}
|
||||
}
|
||||
// Only keep in memcache until the time it would have expired (to avoid clutering memcache)
|
||||
$m->set($sKey, $aCurrentBlock, $t + 1 + $aCurrentBlock[0]/$iLeakPerSecond);
|
||||
}
|
||||
|
||||
// Bucket result in the largest bucket we find
|
||||
$iMaxVal = max($iMaxVal, $aCurrentBlock[0]);
|
||||
}
|
||||
|
||||
return $iMaxVal;
|
||||
}
|
||||
|
||||
function byValue1($a, $b)
|
||||
{
|
||||
if ($a[1] == $b[1])
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
return ($a[1] > $b[1]) ? -1 : 1;
|
||||
}
|
||||
|
||||
function byLastBlockTime($a, $b)
|
||||
{
|
||||
if ($a['lastBlockTimestamp'] == $b['lastBlockTimestamp'])
|
||||
{
|
||||
return 0;
|
||||
}
|
||||
return ($a['lastBlockTimestamp'] > $b['lastBlockTimestamp']) ? -1 : 1;
|
||||
}
|
||||
|
||||
function getBucketBlocks()
|
||||
{
|
||||
$m = getBucketMemcache();
|
||||
if (!$m) return null;
|
||||
$t = time();
|
||||
$aBlockedList = $m->get('blockedList', null, $hCasToken);
|
||||
if (!$aBlockedList) $aBlockedList = array();
|
||||
foreach($aBlockedList as $sKey => $aDetails)
|
||||
{
|
||||
$aCurrentBlock = $m->get($sKey);
|
||||
if (!$aCurrentBlock) $aCurrentBlock = array(0, $t);
|
||||
$iCurrentBucketSize = max(0, $aCurrentBlock[0] - ($t - $aCurrentBlock[1])*CONST_ConnectionBucket_LeakRate);
|
||||
$aBlockedList[$sKey] = array(
|
||||
'totalBlocks' => $aDetails[0],
|
||||
'lastBlockTimestamp' => $aDetails[1],
|
||||
'currentBucketSize' => $iCurrentBucketSize,
|
||||
'currentlyBlocked' => $iCurrentBucketSize + (CONST_ConnectionBucket_Cost_Reverse) >= CONST_ConnectionBucket_BlockLimit,
|
||||
);
|
||||
}
|
||||
uasort($aBlockedList, 'byLastBlockTime');
|
||||
return $aBlockedList;
|
||||
}
|
||||
|
||||
function clearBucketBlocks()
|
||||
{
|
||||
$m = getBucketMemcache();
|
||||
if (!$m) return false;
|
||||
$m->delete('blockedList');
|
||||
return true;
|
||||
}
|
||||
@@ -17,6 +17,7 @@
|
||||
// Connection buckets to rate limit people being nasty
|
||||
@define('CONST_ConnectionBucket_MemcacheServerAddress', false);
|
||||
@define('CONST_ConnectionBucket_MemcacheServerPort', 11211);
|
||||
@define('CONST_ConnectionBucket_MaxBlockList', 100);
|
||||
@define('CONST_ConnectionBucket_LeakRate', 1);
|
||||
@define('CONST_ConnectionBucket_BlockLimit', 10);
|
||||
@define('CONST_ConnectionBucket_WaitLimit', 6);
|
||||
|
||||
@@ -11,6 +11,7 @@
|
||||
array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
|
||||
array('list', 'l', 0, 1, 0, 0, 'bool', 'List recent blocks'),
|
||||
array('delete', 'd', 0, 1, 0, 0, 'bool', 'Clear recent blocks list'),
|
||||
array('flush', '', 0, 1, 0, 0, 'bool', 'Flush all blocks / stats'),
|
||||
);
|
||||
getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
|
||||
|
||||
@@ -28,11 +29,13 @@
|
||||
|
||||
$aBlocks = getBucketBlocks();
|
||||
echo "\n";
|
||||
printf(" %-40s | %12s | %7s | %13s | %16s | %31s\n", "Key", "Total Blocks", "Current", "Still Blocked", "Last Req Blocked", "Last Block Time");
|
||||
printf(" %'--40s-|-%'-12s-|-%'-7s-|-%'-13s-|-%'-16s-|-%'-31s\n", "", "", "", "", "", "");
|
||||
printf(" %-40s | %12s | %7s | %13s | %31s\n", "Key", "Total Blocks", "Current", "Still Blocked", "Last Block Time");
|
||||
printf(" %'--40s-|-%'-12s-|-%'-7s-|-%'-13s-|-%'-31s\n", "", "", "", "", "");
|
||||
foreach($aBlocks as $sKey => $aDetails)
|
||||
{
|
||||
printf(" %-40s | %12s | %7s | %13s | %16s | %31s\n", $sKey, $aDetails['totalBlocks'], (int)$aDetails['currentBucketSize'], $aDetails['lastRequestBlocked']?'Y':'N', $aDetails['currentlyBlocked']?'Y':'N', date("r", $aDetails['lastBlockTimestamp']));
|
||||
printf(" %-40s | %12s | %7s | %13s | %31s\n", $sKey, $aDetails['totalBlocks'],
|
||||
(int)$aDetails['currentBucketSize'], $aDetails['currentlyBlocked']?'Y':'N',
|
||||
date("r", $aDetails['lastBlockTimestamp']));
|
||||
}
|
||||
echo "\n";
|
||||
}
|
||||
@@ -42,3 +45,8 @@
|
||||
$m->set('sleepCounter', 0);
|
||||
clearBucketBlocks();
|
||||
}
|
||||
|
||||
if ($aResult['flush'])
|
||||
{
|
||||
$m->flush();
|
||||
}
|
||||
|
||||
@@ -1,4 +1,6 @@
|
||||
<?php
|
||||
@define('CONST_ConnectionBucket_PageType', 'Details');
|
||||
|
||||
require_once(dirname(dirname(__FILE__)).'/lib/init-website.php');
|
||||
require_once(CONST_BasePath.'/lib/log.php');
|
||||
|
||||
|
||||
@@ -1,4 +1,6 @@
|
||||
<?php
|
||||
@define('CONST_ConnectionBucket_PageType', 'Reverse');
|
||||
|
||||
require_once(dirname(dirname(__FILE__)).'/lib/init-website.php');
|
||||
require_once(CONST_BasePath.'/lib/log.php');
|
||||
|
||||
|
||||
@@ -1,4 +1,6 @@
|
||||
<?php
|
||||
@define('CONST_ConnectionBucket_PageType', 'Search');
|
||||
|
||||
require_once(dirname(dirname(__FILE__)).'/lib/init-website.php');
|
||||
require_once(CONST_BasePath.'/lib/log.php');
|
||||
|
||||
|
||||
Reference in New Issue
Block a user