X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/1461dc603e7d99a56b65057663bcd039b8309d20..2db6f931b6ce1d331433f7e4f02ac332381a7b29:/utils/update.php diff --git a/utils/update.php b/utils/update.php index e26a9538..7448348f 100755 --- a/utils/update.php +++ b/utils/update.php @@ -26,10 +26,12 @@ 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-relation', '', 0, 1, 1, 1, 'int', 'Re-import relation'), + array('import-from-main-api', '', 0, 1, 0, 0, 'bool', 'Use OSM API instead of Overpass to download objects'), array('index', '', 0, 1, 0, 0, 'bool', 'Index'), array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'), @@ -45,7 +47,8 @@ showUsage($aCMDOptions, true, 'Select either import of hourly or daily'); } - if (!isset($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) @@ -65,10 +68,26 @@ // 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); + // 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) @@ -99,7 +118,7 @@ $sNextFile = $aResult['import-diff']; if (!file_exists($sNextFile)) { - echo "Cannot open $nextFile\n"; + echo "Cannot open $sNextFile\n"; exit; } // Don't update the import status - we don't know what this file contains @@ -110,7 +129,7 @@ if (($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); @@ -131,6 +150,7 @@ $bModifyXML = false; $sModifyXMLstr = ''; + $bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api']; if (isset($aResult['import-file']) && $aResult['import-file']) { $bModifyXML = true; @@ -138,23 +158,45 @@ if (isset($aResult['import-node']) && $aResult['import-node']) { $bModifyXML = true; - $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']); + if ($bUseOSMApi) + { + $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']); + } + else + { + $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;'); + } } if (isset($aResult['import-way']) && $aResult['import-way']) { $bModifyXML = true; - $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full'); + if ($bUseOSMApi) + { + $sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full'; + } + else + { + $sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;'; + } + $sModifyXMLstr = file_get_contents($sCmd); } if (isset($aResult['import-relation']) && $aResult['import-relation']) { $bModifyXML = true; - $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full'); + if ($bUseOSMApi) + { + $sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full'); + } + else + { + $sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;'); + } } if ($bModifyXML) { // derive change from normal osm file with osmosis $sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc'; - if ($aResult['import-file']) + if (isset($aResult['import-file']) && $aResult['import-file']) { $sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile; echo $sCMD."\n"; @@ -198,8 +240,7 @@ } // import generated change file - $aPipes = array(); - $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile; + $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile; echo $sCMD."\n"; exec($sCMD, $sJunk, $iErrorLevel); if ($iErrorLevel) @@ -318,18 +359,25 @@ if ($aResult['index']) { - index($aResult, $sDatabaseDSN); + if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1; + passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']); } if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) { + + if (strpos(CONST_Replication_Url, 'download.geofabrik.de') !== false && CONST_Replication_Update_Interval < 86400) { + echo "Error: Update interval too low for download.geofabrik.de. Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n"; + exit; + } + $sImportFile = CONST_BasePath.'/data/osmosischange.osc'; $sOsmosisCMD = CONST_Osmosis_Binary; $sOsmosisConfigDirectory = CONST_BasePath.'/settings'; - $sDatabaseName = 'nominatim'; $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile; - $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$sDatabaseName.' '.$sImportFile; - $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -t '.$aResult['index-instances']; + $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory; + $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile; + $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database']; if (!$aResult['no-npi']) { $sCMDIndex .= '-F '; } @@ -347,11 +395,35 @@ // { if (!file_exists($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, $iErrorLevel); + } + // There are new replication files - use osmosis to download the file + echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n"; + } + $fStartTime = time(); $fCMDStartTime = time(); echo $sCMDDownload."\n"; exec($sCMDDownload, $sJunk, $iErrorLevel); - while ($iErrorLevel == 1) + while ($iErrorLevel > 0) { echo "Error: $iErrorLevel\n"; sleep(60); @@ -360,9 +432,10 @@ } $iFileSize = filesize($sImportFile); $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory); - echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')"; + var_Dump($sSQL); $oDB->query($sSQL); + echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; } $iFileSize = filesize($sImportFile); @@ -375,12 +448,12 @@ if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } - echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')"; var_Dump($sSQL); $oDB->query($sSQL); + echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; // Archive for debug? unlink($sImportFile); @@ -389,16 +462,25 @@ $sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory); // Index file - $sThisIndexCmd = $sCMDIndex; + if (!isset($aResult['index-instances'])) + { + if (getLoadAverage() < 15) + $iIndexInstances = 2; + else + $iIndexInstances = 1; + } else + $iIndexInstances = $aResult['index-instances']; + + $sThisIndexCmd = $sCMDIndex.' -t '.$iIndexInstances; + $fCMDStartTime = time(); if (!$aResult['no-npi']) { - $fCMDStartTime = time(); $iFileID = $oDB->getOne('select nextval(\'file\')'); if (PEAR::isError($iFileID)) { echo $iFileID->getMessage()."\n"; - exit; + exit(-1); } $sFileDir = CONST_BasePath.'/export/diff/'; $sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT); @@ -424,7 +506,7 @@ if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } if (!$aResult['no-npi']) @@ -439,7 +521,7 @@ if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2", @@ -447,19 +529,28 @@ } } - echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; $sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')"; + var_Dump($sSQL); $oDB->query($sSQL); + echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n"; $sSQL = "update import_status set lastimportdate = '$sBatchEnd'"; $oDB->query($sSQL); $fDuration = time() - $fStartTime; - echo "Completed for $sBatchEnd in ".round($fDuration/60,2)."\n"; + echo date('Y-m-d H:i:s')." Completed all for $sBatchEnd in ".round($fDuration/60,2)." minutes\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 ) + { + $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time())); + } + else + { + $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration); + } + echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n"; + sleep($iSleep); } }