allow osm2pgsql cache memory to be set for updates

[nominatim.git] / utils / update.php

diff --git a/utils/update.php b/utils/update.php
index eadd385136c1f22b9d87ccfcc3a732181691c2d4..93b5ab0919db374dab7da461f830c19d451985bc 100755 (executable)
--- a/utils/update.php
+++ b/utils/update.php
@@ -26,6 +26,7 @@
 
                array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
                array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),
 
                array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
                array('import-diff', '', 0, 1, 1, 1, 'realpath', 'Import a diff (osc) file from local file system'),

+               array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),

 
                array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
                array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),
 
                array('import-node', '', 0, 1, 1, 1, 'int', 'Re-import node'),
                array('import-way', '', 0, 1, 1, 1, 'int', 'Re-import way'),


@@ -48,6 +49,7 @@
 
        if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
        if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
 
        if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
        if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;

+

 /*
        // Lock to prevent multiple copies running
        if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
 /*
        // Lock to prevent multiple copies running
        if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)


@@ -67,10 +69,26 @@
        // Assume osm2pgsql is in the folder above
        $sBasePath = dirname(dirname(__FILE__));
 
        // Assume osm2pgsql is in the folder above
        $sBasePath = dirname(dirname(__FILE__));
 

+       date_default_timezone_set('Etc/UTC');
+

        $oDB =& getDB();
 
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 
        $oDB =& getDB();
 
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 

+       // cache memory to be used by osm2pgsql, should not be more than the available memory
+       $iCacheMemory = (isset($aResult['osm2pgsql-cache'])?$aResult['osm2pgsql-cache']:2000);
+       if ($iCacheMemory + 500 > getTotalMemoryMB())
+       {
+               $iCacheMemory = getCacheMemoryMB();
+               echo "WARNING: resetting cache memory to $iCacheMemory\n";
+       }
+       $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'];
+       if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+       {
+               $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+       }
+
+

        $bFirst = true;
        $bContinue = $aResult['import-all'];
        while ($bContinue || $bFirst)
        $bFirst = true;
        $bContinue = $aResult['import-all'];
        while ($bContinue || $bFirst)


@@ -101,7 +119,7 @@
                        $sNextFile = $aResult['import-diff'];
                        if (!file_exists($sNextFile))
                        {
                        $sNextFile = $aResult['import-diff'];
                        if (!file_exists($sNextFile))
                        {

-                               echo "Cannot open $nextFile\n";
+                               echo "Cannot open $sNextFile\n";

                                exit;
                        }
                        // Don't update the import status - we don't know what this file contains
                                exit;
                        }
                        // Don't update the import status - we don't know what this file contains


@@ -112,7 +130,7 @@
                if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
                {
                        // Import the file
                if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
                {
                        // Import the file

-                       $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sNextFile;
+                       $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;

                        echo $sCMD."\n";
                        exec($sCMD, $sJunk, $iErrorLevel);
 
                        echo $sCMD."\n";
                        exec($sCMD, $sJunk, $iErrorLevel);
 


@@ -147,7 +165,7 @@
                }
                else
                {
                }
                else
                {

-                       $sModifyXMLstr = file_get_contents('http://overpass.osm.rambler.ru/cgi/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
+                       $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');

                }
        }
        if (isset($aResult['import-way']) && $aResult['import-way'])
                }
        }
        if (isset($aResult['import-way']) && $aResult['import-way'])


@@ -159,7 +177,7 @@
                }
                else
                {
                }
                else
                {

-                       $sCmd = 'http://overpass.osm.rambler.ru/cgi/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
+                       $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';

                }
                $sModifyXMLstr = file_get_contents($sCmd);
        }
                }
                $sModifyXMLstr = file_get_contents($sCmd);
        }


@@ -172,7 +190,7 @@
                }
                else
                {
                }
                else
                {

-                       $sModifyXMLstr = file_get_contents('http://overpass.osm.rambler.ru/cgi/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
+                       $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');

                }
        }
        if ($bModifyXML)
                }
        }
        if ($bModifyXML)


@@ -223,7 +241,7 @@
                }
 
                // import generated change file
                }
 
                // import generated change file

-               $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile;
+               $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;

                echo $sCMD."\n";
                exec($sCMD, $sJunk, $iErrorLevel);
                if ($iErrorLevel)
                echo $sCMD."\n";
                exec($sCMD, $sJunk, $iErrorLevel);
                if ($iErrorLevel)


@@ -351,7 +369,8 @@
                $sOsmosisCMD = CONST_Osmosis_Binary;
                $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
                $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
                $sOsmosisCMD = CONST_Osmosis_Binary;
                $sOsmosisConfigDirectory = CONST_BasePath.'/settings';
                $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;

-               $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sImportFile;
+               $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
+               $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;

                $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'];
                if (!$aResult['no-npi']) {
                        $sCMDIndex .= '-F ';
                $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'];
                if (!$aResult['no-npi']) {
                        $sCMDIndex .= '-F ';


@@ -370,11 +389,34 @@
 //                     {
                                if (!file_exists($sImportFile))
                                {
 //                     {
                                if (!file_exists($sImportFile))
                                {

-                                       // Use osmosis to download the file
+                                       // First check if there are new updates published (except for minutelies - there's always new diffs to process)
+                                       if ( CONST_Replication_Update_Interval > 60 )
+                                       {
+
+                                               unset($aReplicationLag);
+                                               exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
+                                               while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
+                                               {
+                                                       if ($iErrorLevel)
+                                                       {
+                                                               echo "Error: $iErrorLevel. ";
+                                                               echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
+                                                       }
+                                                       else
+                                                       {
+                                                               echo ".";
+                                                       }
+                                                       sleep(CONST_Replication_Recheck_Interval);
+                                                       unset($aReplicationLag);
+                                                       exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); 
+                                               }
+                                               // There are new replication files - use osmosis to download the file
+                                               echo "\nReplication Delay is ".$aReplicationLag[0]."\n";
+                                       }

                                        $fCMDStartTime = time();
                                        echo $sCMDDownload."\n";
                                        exec($sCMDDownload, $sJunk, $iErrorLevel);
                                        $fCMDStartTime = time();
                                        echo $sCMDDownload."\n";
                                        exec($sCMDDownload, $sJunk, $iErrorLevel);

-                                       while ($iErrorLevel == 1)
+                                       while ($iErrorLevel > 0)

                                        {
                                                echo "Error: $iErrorLevel\n";
                                                sleep(60);
                                        {
                                                echo "Error: $iErrorLevel\n";
                                                sleep(60);


@@ -481,8 +523,16 @@
                        echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
                        if (!$aResult['import-osmosis-all']) exit;
 
                        echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n";
                        if (!$aResult['import-osmosis-all']) exit;
 

-                       echo "Sleeping ".max(0,60-$fDuration)." seconds\n";
-                       sleep(max(0,60-$fDuration));
+                       if ( CONST_Replication_Update_Interval > 60 )
+                       {
+                               $iSleep = round(CONST_Replication_Update_Interval*0.8);
+                       }
+                       else
+                       {
+                               $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration);
+                       }
+                       echo "Sleeping $iSleep seconds\n";
+                       sleep($iSleep);

                }
 
        }
                }
 
        }