Do not truncate aggregate table

ubccr · Mar 15, 2019 · 55aed11 · 55aed11
1 parent 90a42f7
commit 55aed11
Show file tree

Hide file tree

Showing 11 changed files with 276 additions and 19 deletions.
diff --git a/bin/xdmod-ingestor b/bin/xdmod-ingestor
@@ -263,11 +263,11 @@ function main()
                 }
 
                 if($datatypeValue == 'openstack'){
-                    $dwi->ingestCloudDataOpenStack();
+                    $dwi->ingestCloudDataOpenStack($lastModifiedStartDate);
                 }
 
                 if($datatypeValue == 'genericcloud'){
-                    $dwi->ingestCloudDataGeneric();
+                    $dwi->ingestCloudDataGeneric($lastModifiedStartDate);
                 }
 
                 if ($datatypeValue == 'storage') {
@@ -293,7 +293,7 @@ function main()
             }
 
             if($realmToAggregate == 'cloud' || $realmToAggregate === false){
-                $dwi->aggregateCloudData();
+                $dwi->aggregateCloudData($lastModifiedStartDate);
             }
 
             if ($realmToAggregate == 'storage' || $realmToAggregate === false) {

diff --git a/classes/ETL/Ingestor/CloudStateReconstructorTransformIngestor.php b/classes/ETL/Ingestor/CloudStateReconstructorTransformIngestor.php
@@ -58,7 +58,7 @@ public function __construct(aOptions $options, EtlConfiguration $etlConfig, Log
         $this->_start_event_ids = array(self::START, self::RESUME, self::STATE_REPORT, self::UNSHELVE, self::UNPAUSE, self::UNSUSPEND, self::POWER_ON);
         $this->_all_event_ids = array_merge($this->_start_event_ids, $this->_stop_event_ids);
         $this->_end_time = $etlConfig->getVariableStore()->endDate ? date('Y-m-d H:i:s', strtotime($etlConfig->getVariableStore()->endDate)) : null;
-
+        
         $this->resetInstance();
     }
 
@@ -132,8 +132,9 @@ protected function getSourceQueryString()
         // is lost. To work around this we add a dummy row filled with zeroes.
         $colCount = count($this->etlSourceQuery->records);
         $unionValues = array_fill(0, $colCount, 0);
+        $subSelect = "(SELECT DISTINCT instance_id from modw_cloud.event WHERE last_modified > \"" . $this->getEtlOverseerOptions()->getLastModifiedStartDate() . "\")";
 
-        $sql = "$sql WHERE event_type_id IN (" . implode(',', $this->_all_event_ids) . ")\nUNION ALL\nSELECT " . implode(',', $unionValues) . "\nORDER BY 1 DESC, 2 DESC, 3 ASC, 4 DESC";
+        $sql = "$sql WHERE instance_id IN " . $subSelect . " AND event_type_id IN (" . implode(',', $this->_all_event_ids) . ")\nUNION ALL\nSELECT " . implode(',', $unionValues) . "\nORDER BY 1 DESC, 2 DESC, 3 ASC, 4 DESC";
 
         return $sql;
     }

diff --git a/classes/OpenXdmod/DataWarehouseInitializer.php b/classes/OpenXdmod/DataWarehouseInitializer.php
@@ -202,12 +202,16 @@ public function ingestAllHpcdb($startDate = null, $endDate = null)
      * tables do not exist then catch the resulting exception and display a message
      * saying that there is no OpenStack data to ingest.
      */
-    public function ingestCloudDataOpenStack()
+    public function ingestCloudDataOpenStack($lastModifiedStartDate)
     {
         if( $this->isRealmEnabled('Cloud') ){
-            try{
+            try {
                 $this->logger->notice('Ingesting OpenStack event log data');
-                Utilities::runEtlPipeline(array('jobs-cloud-import-users-openstack', 'jobs-cloud-extract-openstack'), $this->logger);
+                Utilities::runEtlPipeline(
+                    array('jobs-cloud-import-users-openstack', 'jobs-cloud-extract-openstack'),
+                    $this->logger,
+                    array('last-modified-start-date' => $lastModifiedStartDate)
+                );
             }
             catch( Exception $e ){
                 if( $e->getCode() == 1146 ){
@@ -225,12 +229,16 @@ public function ingestCloudDataOpenStack()
      * tables do not exist then catch the resulting exception and display a message
      * saying that there is no generic cloud data to ingest.
      */
-    public function ingestCloudDataGeneric()
+    public function ingestCloudDataGeneric($lastModifiedStartDate)
     {
         if( $this->isRealmEnabled('Cloud') ){
-            try{
+            try {
                 $this->logger->notice('Ingesting generic cloud log files');
-                Utilities::runEtlPipeline(array('jobs-cloud-import-users-generic', 'jobs-cloud-extract-generic'), $this->logger);
+                Utilities::runEtlPipeline(
+                    array('jobs-cloud-import-users-generic', 'jobs-cloud-extract-generic'),
+                    $this->logger,
+                    array('last-modified-start-date' => $lastModifiedStartDate)
+                );
             }
             catch( Exception $e ){
                 if( $e->getCode() == 1146 ){
@@ -273,11 +281,15 @@ public function ingestStorageData()
      * catch the resulting exception and display a message saying that there
      * is no cloud data to aggregate and cloud aggregation is being skipped.
      */
-    public function aggregateCloudData()
+    public function aggregateCloudData($lastModifiedStartDate)
     {
         if( $this->isRealmEnabled('Cloud') ){
             $this->logger->notice('Aggregating Cloud data');
-            Utilities::runEtlPipeline(array('cloud-state-pipeline'), $this->logger);
+            Utilities::runEtlPipeline(
+                array('cloud-state-pipeline'),
+                $this->logger,
+                array('last-modified-start-date' => $lastModifiedStartDate)
+            );
 
             $filterListBuilder = new FilterListBuilder();
             $filterListBuilder->setLogger($this->logger);

diff --git a/configuration/etl/etl.d/cloud_state_machine.json b/configuration/etl/etl.d/cloud_state_machine.json
@@ -35,7 +35,6 @@
             "name": "cloud-transient",
             "class": "DatabaseIngestor",
             "definition_file": "cloud_common/cloud_transient.json",
-            "truncate_destination": true,
             "description": "Builds intermediate cloud event table"
         },
         {
@@ -46,7 +45,6 @@
             "class": "SimpleAggregator",
             "description": "Aggregate cloud records.",
             "definition_file": "cloud_common/cloud_metrics_aggregation.json",
-            "truncate_destination": true,
             "table_prefix": "cloudfact_by_",
             "aggregation_units": [
                 "day", "month", "quarter", "year"

diff --git a/configuration/etl/etl_action_defs.d/cloud_common/cloud_metrics_aggregation.json b/configuration/etl/etl_action_defs.d/cloud_common/cloud_metrics_aggregation.json
@@ -5,8 +5,8 @@
         },
         "aggregation_period_query": {
             "overseer_restrictions": {
-                "#last_modified_start_date": "last_modified >= ${VALUE}",
-                "#last_modified_end_date": "last_modified <= ${VALUE}",
+                "last_modified_start_date": "last_modified >= ${VALUE}",
+                "last_modified_end_date": "last_modified <= ${VALUE}",
                 "include_only_resource_codes": "resource_id IN ${VALUE}",
                 "exclude_resource_codes": "resource_id NOT IN ${VALUE}"
             }

diff --git a/configuration/etl/etl_tables.d/cloud_common/cloud_transient.json b/configuration/etl/etl_tables.d/cloud_common/cloud_transient.json
@@ -103,6 +103,13 @@
                 "name": "submission_venue_id",
                 "type": "int(5)",
                 "nullable": true
+            },
+            {
+                "name": "last_modified",
+                "type": "timestamp",
+                "nullable": false,
+                "default": "CURRENT_TIMESTAMP",
+                "extra": "on update CURRENT_TIMESTAMP"
             }
         ],
         "indexes": [
@@ -111,7 +118,7 @@
                 "columns": [
                     "resource_id",
                     "instance",
-                    "start_time"
+                    "start_time_ts"
                 ],
                 "is_unique": true
             },
@@ -120,9 +127,15 @@
                 "columns": [
                     "instance_id",
                     "resource_id",
-                    "start_time"
+                    "start_time_ts"
                 ],
                 "is_unique": true
+            },
+            {
+                "name": "index_last_modified",
+                "columns": [
+                    "last_modified"
+                ]
             }
         ]
     }

diff --git a/configuration/etl/etl_tables.d/cloud_common/event.json b/configuration/etl/etl_tables.d/cloud_common/event.json
@@ -68,6 +68,13 @@
                 "type": "int(5)",
                 "nullable": false,
                 "default": -1
+            },
+            {
+                "name": "last_modified",
+                "type": "timestamp",
+                "nullable": false,
+                "default": "CURRENT_TIMESTAMP",
+                "extra": "on update CURRENT_TIMESTAMP"
             }
         ],
         "indexes": [
@@ -110,6 +117,12 @@
                     "resource_id"
                 ],
                 "is_unique": false
+            },
+            {
+                "name": "index_last_modified",
+                "columns": [
+                    "last_modified"
+                ]
             }
         ]
     }

diff --git a/...xdmod/modules/xdmod/regression_tests/lib/Controllers/UsageExplorerCloudPostIngestTest.php b/...xdmod/modules/xdmod/regression_tests/lib/Controllers/UsageExplorerCloudPostIngestTest.php
@@ -0,0 +1,32 @@
+<?php
+
+namespace Controllers;
+
+class UsageExplorerCloudPostIngestTest extends UsageExplorerTest
+{
+
+    public function csvExportProvider(){
+        parent::$baseDir = __DIR__ . '/../../../tests/artifacts/xdmod-test-artifacts/xdmod/post_ingest/';
+        parent::defaultSetup();
+
+        $statistics = array(
+            'cloud_num_sessions_ended',
+            'cloud_num_sessions_running',
+            'cloud_num_sessions_started',
+        );
+
+        $group_bys = array(
+            'none',
+        );
+
+        $varSettings = array(
+            'realm' => array('Cloud'),
+            'dataset_type' => array('aggregate', 'timeseries'),
+            'statistic' => $statistics,
+            'group_by' => $group_bys,
+            'aggregation_unit' => array_keys($this->aggregationUnits)
+        );
+
+        return parent::generateTests($varSettings, '2018-05-19', '2018-05-19');
+    }
+}
diff --git a/open_xdmod/modules/xdmod/regression_tests/post_ingest_test.sh b/open_xdmod/modules/xdmod/regression_tests/post_ingest_test.sh
@@ -0,0 +1,36 @@
+#!/bin/sh
+# Tests that cloud ETL works after upgrade + reingestion
+
+BASEDIR=./open_xdmod/modules/xdmod/regression_tests
+REF_DIR=/var/tmp/referencedata
+last_modified_start_date=$(date +'%F %T')
+
+oldCount=$(mysql -N -B -e "SELECT SUM(num_sessions_ended) FROM modw_cloud.cloudfact_by_month WHERE month = 4;")
+
+if [ $oldCount -ne 53 ]
+then
+    echo " Count $oldCount did not match expected result of 53"
+    exit 1
+fi
+
+sudo -u xdmod xdmod-shredder -r openstack -d $REF_DIR/openstack_upgrade -f openstack
+sudo -u xdmod xdmod-ingestor --datatype=openstack --last-modified-start-date "$last_modified_start_date"
+sudo -u xdmod xdmod-ingestor --aggregate=cloud --last-modified-start-date "$last_modified_start_date"
+
+newCount=$(mysql -N -B -e "SELECT SUM(num_sessions_ended) FROM modw_cloud.cloudfact_by_month WHERE month = 4;")
+
+if [ $newCount -ne 52 ]
+then
+    echo " Count $newCount did not match expected result of 52"
+    exit 1
+
+fi
+
+newRows=$(mysql -N -B -e "SELECT SUM(num_sessions_ended) FROM modw_cloud.cloudfact_by_month WHERE month = 5;")
+
+if [ $newRows -ne 2 ]
+then
+    echo " Count $newRows did not match expected result of 2"
+    exit 1
+
+fi
diff --git a/shippable.yml b/shippable.yml
@@ -25,6 +25,7 @@ build:
         - cp ./configuration/portal_settings.ini ./configuration/portal_settings.ini.old
         - cp -f /etc/xdmod/portal_settings.ini ./configuration/portal_settings.ini
         - ./open_xdmod/modules/xdmod/integration_tests/runtests.sh --junit-output-dir `pwd`/shippable/testresults/
+        - ./open_xdmod/modules/xdmod/regression_tests/post_ingest_test.sh --junit-output-dir `pwd`/shippable/testresults/
         - ./open_xdmod/modules/xdmod/component_tests/runtests.sh --log-junit `pwd`/shippable/testresults/xdmod-component.xml
         - ./open_xdmod/modules/xdmod/automated_tests/runtests.sh --headless --log-junit `pwd`/shippable/testresults
         - ./open_xdmod/modules/xdmod/integration_tests/scripts/samlSetup.sh