X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/8fd7d319b7ef6ee4d4f8cf3b61eb47903e7fb1eb..af150bc63eb6893ea06577a1534beb85a5dbf29e:/utils/update.php diff --git a/utils/update.php b/utils/update.php index 0e26b311..cc275463 100755 --- 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('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'), @@ -48,19 +49,18 @@ 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, $iResult) > 1) { - echo "Copy already running\n"; - exit; + fail("Copy already running\n"); } if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9; if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3; if (getBlockingProcesses() > $aResult['max-blocking']) { - echo "Too many blocking processes for import\n"; - exit; + fail("Too many blocking processes for import\n"); } */ @@ -72,6 +72,21 @@ $oDB =& getDB(); $aDSNInfo = DB::parseDSN(CONST_Database_DSN); + if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432; + + // 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"; + } + $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port']; + if (!is_null(CONST_Osm2pgsql_Flatnode_File)) + { + $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; + } + $bFirst = true; $bContinue = $aResult['import-all']; @@ -103,8 +118,7 @@ $sNextFile = $aResult['import-diff']; if (!file_exists($sNextFile)) { - echo "Cannot open $sNextFile\n"; - exit; + fail("Cannot open $sNextFile\n"); } // Don't update the import status - we don't know what this file contains $sUpdateSQL = 'update import_status set lastimportdate = now() where false'; @@ -114,18 +128,13 @@ if (($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; - if (!is_null(CONST_Osm2pgsql_Flatnode_File)) - { - $sCMD .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; - } + $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile; echo $sCMD."\n"; exec($sCMD, $sJunk, $iErrorLevel); if ($iErrorLevel) { - echo "Error from osm2pgsql, $iErrorLevel\n"; - exit; + fail("Error from osm2pgsql, $iErrorLevel\n"); } // Move the date onwards @@ -192,8 +201,7 @@ exec($sCMD, $sJunk, $iErrorLevel); if ($iErrorLevel) { - echo "Error converting osm to osc, osmosis returned: $iErrorLevel\n"; - exit; + fail("Error converting osm to osc, osmosis returned: $iErrorLevel\n"); } } else @@ -208,8 +216,7 @@ $hProc = proc_open($sCMD, $aSpec, $aPipes); if (!is_resource($hProc)) { - echo "Error converting osm to osc, osmosis failed\n"; - exit; + fail("Error converting osm to osc, osmosis failed\n"); } fwrite($aPipes[0], $sModifyXMLstr); fclose($aPipes[0]); @@ -224,27 +231,28 @@ echo "Error converting osm to osc, osmosis returned: $iError\n"; echo $sOut; echo $sErrors; - exit; + exit(-1); } } // import generated change file - $sCMD = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sTemporaryFile; - if (!is_null(CONST_Osm2pgsql_Flatnode_File)) - { - $sCMD .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; - } + $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile; echo $sCMD."\n"; exec($sCMD, $sJunk, $iErrorLevel); if ($iErrorLevel) { - echo "osm2pgsql exited with error level $iErrorLevel\n"; - exit; + fail("osm2pgsql exited with error level $iErrorLevel\n"); } } if ($aResult['deduplicate']) { + + $pgver = (float) CONST_Postgresql_Version; + if ($pgver < 9.3) { + fail("ERROR: deduplicate is only currently supported in postgresql 9.3"); + } + $oDB =& getDB(); $sSQL = 'select partition from country_name order by country_code'; $aPartitions = $oDB->getCol($sSQL); @@ -265,76 +273,65 @@ if (PEAR::isError($aTokenSet)) { var_dump($aTokenSet, $sSQL); - exit; + exit(1); } - + $aKeep = array_shift($aTokenSet); $iKeepID = $aKeep['word_id']; foreach($aTokenSet as $aRemove) { $sSQL = "update search_name set"; - $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.","; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID."),"; + $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) { var_dump($x); - exit; + exit(1); } $sSQL = "update search_name set"; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " nameaddress_vector = array_replace(nameaddress_vector,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) { var_dump($x); - exit; + exit(1); } $sSQL = "update location_area_country set"; - $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) { var_dump($x); - exit; + exit(1); } foreach ($aPartitions as $sPartition) { $sSQL = "update search_name_".$sPartition." set"; - $sSQL .= " name_vector = (name_vector - ".$aRemove['word_id'].")+".$iKeepID.","; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " name_vector = array_replace(name_vector,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where name_vector @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) { var_dump($x); - exit; - } - - $sSQL = "update search_name_".$sPartition." set"; - $sSQL .= " nameaddress_vector = (nameaddress_vector - ".$aRemove['word_id'].")+".$iKeepID; - $sSQL .= " where nameaddress_vector @> ARRAY[".$aRemove['word_id']."]"; - $x = $oDB->query($sSQL); - if (PEAR::isError($x)) - { - var_dump($x); - exit; + exit(1); } $sSQL = "update location_area_country set"; - $sSQL .= " keywords = (keywords - ".$aRemove['word_id'].")+".$iKeepID; + $sSQL .= " keywords = array_replace(keywords,".$aRemove['word_id'].",".$iKeepID.")"; $sSQL .= " where keywords @> ARRAY[".$aRemove['word_id']."]"; $x = $oDB->query($sSQL); if (PEAR::isError($x)) { var_dump($x); - exit; + exit(1); } } @@ -343,7 +340,7 @@ if (PEAR::isError($x)) { var_dump($x); - exit; + exit(1); } } @@ -352,22 +349,23 @@ if ($aResult['index']) { - passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']); + passthru(CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -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) { + fail("Error: Update interval too low for download.geofabrik.de. Please check install documentation (http://wiki.openstreetmap.org/wiki/Nominatim/Installation#Updates)\n"); + } + $sImportFile = CONST_BasePath.'/data/osmosischange.osc'; $sOsmosisCMD = CONST_Osmosis_Binary; $sOsmosisConfigDirectory = CONST_BasePath.'/settings'; $sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile; $sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory; - $sCMDImport = CONST_Osm2pgsql_Binary.' -klas -C 2000 -O gazetteer -d '.$aDSNInfo['database'].' '.$sImportFile; - if (!is_null(CONST_Osm2pgsql_Flatnode_File)) - { - $sCMDImport .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File; - } - $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$aResult['index-instances']; + $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile; + $sCMDIndex = $sBasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances']; if (!$aResult['no-npi']) { $sCMDIndex .= '-F '; } @@ -407,8 +405,9 @@ exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel); } // There are new replication files - use osmosis to download the file - echo "\nReplication Delay is ".$aReplicationLag[0]."\n"; + 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); @@ -421,9 +420,10 @@ } $iFileSize = filesize($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); @@ -436,12 +436,12 @@ 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); @@ -451,15 +451,15 @@ // Index file $sThisIndexCmd = $sCMDIndex; + $fCMDStartTime = time(); if (!$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); @@ -485,7 +485,7 @@ if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } if (!$aResult['no-npi']) @@ -500,7 +500,7 @@ if ($iErrorLevel) { echo "Error: $iErrorLevel\n"; - exit; + exit($iErrorLevel); } rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2", @@ -508,26 +508,27 @@ } } - 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"; - if (!$aResult['import-osmosis-all']) exit; + 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(0); if ( CONST_Replication_Update_Interval > 60 ) { - $iSleep = round(CONST_Replication_Update_Interval*0.8); + $iSleep = max(0,(strtotime($sBatchEnd)+CONST_Replication_Update_Interval-time())); } else { $iSleep = max(0,CONST_Replication_Update_Interval-$fDuration); } - echo "Sleeping $iSleep seconds\n"; + echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n"; sleep($iSleep); } @@ -539,7 +540,7 @@ if (PEAR::isError($iNPIID)) { var_dump($iNPIID); - exit; + exit(1); } $sConfigDirectory = CONST_BasePath.'/settings'; $sCMDImportTemplate = $sBasePath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.$sBasePath.'/nominatim/partitionedtags.def -F '; @@ -570,8 +571,7 @@ exec($sCMDImport, $sJunk, $iErrorLevel); if ($iErrorLevel) { - echo "Error: $iErrorLevel\n"; - exit; + fail("Error: $iErrorLevel\n"); } $sBatchEnd = $iNPIID; echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";