Update to latest Quepid... Requires Mysql8, and removes RRE (#163)

* quepid now wants mysql 8, but smui is 5 I think. 

* workaround to make SMUI work with mysql 8...

* RRE won't run due to Maven not finding some jars so removed :-(.

.gitignore

@@ -4,7 +4,6 @@ volumes/solr*
 solr/configsets/ecommerce.zip
 solr/data/*.json
 solr/data/*.gz
-rre/target
 .idea
 *.iml
 __pycache__

README.md

@@ -41,7 +41,6 @@ We host a complete demonstration environment online for you to play with, see li
 * Solr runs at http://localhost:8983 |  http://chorus.dev.o19s.com:8983
 * SMUI runs at http://localhost:9000 |  http://chorus.dev.o19s.com:9000
 * Quepid runs at http://localhost:3000 |  http://chorus.dev.o19s.com:3000
-* RRE runs at http://localhost:7979 |  http://chorus.dev.o19s.com:7979
 * Keycloak runs at http://keycloak:9080 |  http://chorus.dev.o19s.com:9080
 * Prometheus runs at http://localhost:9090 |  http://chorus.dev.o19s.com:9090
 * Grafana runs at http://localhost:9091 |  http://chorus.dev.o19s.com:9091

TECHNICAL_DETAILS.md

@@ -12,7 +12,7 @@
 | MySQL     | root                        | password  | docker-compose.yml |
 
 
-`blockUnknown` is false as we want to let RRE run against the `ecommerce` collection.  We have locked down in `security.json` to allow anonymous users only to hit the /ecommerce/select/ end point
+`blockUnknown` is false as we want to let RRE (now removed from Chorus) run against the `ecommerce` collection.  We have locked down in `security.json` to allow anonymous users only to hit the /ecommerce/select/ end point
 
 When you bring up Solr Admin and then are redirected to Keycloak, when you register and sent back to Solr you are given the `solr-admin` role.
 
@@ -48,13 +48,6 @@ docker-compose run quepid thor user:create -a demo@example.com "Demo User" passw
 
 For Quepid case, pick `name` for title, and `id` for identifier.  Add `thumb:imageUrl` to the list of fields.
 
-## Setting up RRE
-
-```sh
-docker-compose run rre mvn rre:evaluate
-docker-compose run rre mvn rre-report:report
-```
-
 ## Monitoring Details
 
 Prometheus and Grafana setup heavily inspired by https://github.com/vegasbrianc/prometheus and https://github.com/chatman/solr-grafana-docker

docker-compose.yml

@@ -22,7 +22,7 @@ services:
 
   mysql:
     container_name: mysql
-    image: amd64/mysql:5.7.40
+    image: mysql:8.3.0
     ports:
       - 3306:3306
     environment:
@@ -173,7 +173,7 @@ services:
 
   quepid:
     container_name: quepid
-    image: o19s/quepid:6.14.0
+    image: o19s/quepid:7.15.1
     ports:
       - 3000:3000
     command: "foreman s -f Procfile"
@@ -205,22 +205,6 @@ services:
       - mysql
       - redis
 
-  rre:
-    container_name: rre
-    build:
-      context: ./rre
-      dockerfile: Dockerfile
-    volumes:
-      - ./rre:/rre
-    ports:
-      - "7979:8080"
-    command: "java -jar rre-server-1.1.jar"
-    environment:
-      - SOLR_URL=http://solr1:8983/solr/ecommerce
-    depends_on:
-      - solr1
-
-
   embeddings:
     container_name: embeddings
     build:

katas/000_setting_up_chorus.md

@@ -152,23 +152,8 @@ Visit Quepid at http://localhost:3000 and log in with the email and password you
 
 Go through the initial case setup process.  Quepid will walk you through setting up a `Movie Cases` case via a Wizard interface, and then show you some of the key features of Quepid's UI.  I know you want to skip the tour of Quepid interface, however there is a lot of interactivity in the UI, so it's worth going through the tutorial to get acquainted!
 
-Now we are ready to confirm our second Offline Testing tool, Rated Ranking Evaluator, commonly called RRE, is ready to go.  Unlike Quepid, which is a webapp, RRE is a set of command line tools that run tests, and then publishes the results in both a Excel spreadsheet format and a web dashboard.   
-
-Now, lets confirm that you can run the RRE command line tool.  Go ahead and run a regression:  
-
-```sh
-docker-compose run rre mvn rre:evaluate
-```
-
-You should see some output, and you should see the output saved to `./rre/target/rre/evaluation.json` in your local directory.  We've wrapped RRE inside of the Docker container, so you can edit the RRE configurations locally, but run RRE in the container.
-
 Now, let's go ahead and make sure we publish the results of our evaluation:
 
-```sh
-docker-compose run rre mvn rre-report:report
-```
-
-You can now see a html page saved to `./rre/target/site/rre-report.html`.  
 
 Bring up http://localhost:7979 and you will see a relatively unexciting empty dashboard.  Don't worry, in our first kata, we'll do a relevancy test and fill this dashboard in.
 

katas/001_optimize_a_query.md

@@ -90,32 +90,6 @@ Notice that our results have now turned green across the board from red?  Our gr
 
 There is a lot to unpack in here that is beyond the scope of this kata.  However bear with me.
 
-So now we feel like this is good results.  However, while we've just tuned these notebook examples, what has the impact of the Querqy rule changes been on potentially other queries?  Quepid is great for up to 60 or so queries per case, but it doesn't handle 100's or 1000's of queries well.  Plus, we only get one search metric, in our case NDCG at a time. What if I want to calculate all the search metrics.  For that, enter RRE.
-
-RRE uses it's own format of ratings, so in Quepid, click the `Export` option in the toolbar.  This will pop up a modal box, and choose the `Rated Ranking Evaluator` option.  Then click `Export` and save this file on your desktop.  
-
-You then need to put this ratings into RRE.  First move any ratings file in `./rre/src/etc/ratings/` out of that directory, leaving them in `./rre` is fine.  Then take your freshly exported `Notebook_Computers_rre.json` file and put it in the `ratings` directory.  
-
-Now, lets look at the RRE setup. We want to regression test our pre Querqy algorithm and our post Querqy algorithm.  So look at the template files in `./rre/src/etc/templates`.  As you can see `v1.0` is just the simplistic query.   However `v1.1` is using the Querqy enabled request handler.  This is going to let us measure the two against each other.
-
-While in this kata we are only running the same set of queries as in Quepid, in real life you would be regression testing those two queries against 100's of other rated queries as well.
-
-So let's go run our evaluation!  We're back on the command line:
-
-```sh
-docker-compose run rre mvn rre:evaluate
-```
-
-Look for a message about `completed all 2 evaluations` to know that it's running properly.
-
-And once that completes, lets go ahead and publish the reports:
-
-```sh
-docker-compose run rre mvn rre-report:report
-```
-
-You have two ways of looking at the results of RRE running.  There is an Excel spreadsheet you can look at, or you can use the online Dashboard available at http://localhost:7979.
-
-Going into what all the metrics that RRE provides, and this is just a small sample set, is beyond this.   Suffice to say, if you look at the NDCG@4 and NDCG@10, you will see that we had a big jump from the terrible results of v1.0 to the amazing results in v1.1!
+So now we feel like this is good results.  However, while we've just tuned these notebook examples, what has the impact of the Querqy rule changes been on potentially other queries?  Quepid is great for up to 60 or so queries per case, but it doesn't handle 100's or 1000's of queries well.  Plus, we only get one search metric, in our case NDCG at a time. 
 
 That all folks!  You've successfully taken a bad 'notebook' query, assessed it to put a numerical value on the quality of the search, and then improved it using some rules to rewrite the query.  You then remeasured, saw the quantitative improvement, and then ran a simulated regression test of those queries (and all your other ones in the real world), and have meaningfully improved search quality, which drives more revenue!

quickstart.sh

@@ -54,9 +54,6 @@ done
 # Function check_prerequisites makes sure that you have curl, jq, docker-compose, and zip installed. See helpers.sh for details.
 check_prerequisites
 
-services="blacklight solr1 solr2 solr3 keycloak smui"
-
-
 services="blacklight solr1 solr2 solr3 keycloak"
 
 if $observability; then
@@ -68,7 +65,7 @@ else
 fi
 
 if $offline_lab; then
-  services="${services} quepid rre"
+  services="${services} quepid"
 fi
 
 if $vector_search; then
@@ -81,9 +78,7 @@ if $vector_search; then
   fi
 fi
 
-if $active_search_management; then
-  services="${services} smui"
-fi
+
 
 if ! $local_deploy; then
   echo -e "${MAJOR}Updating configuration files for online deploy${RESET}"
@@ -101,6 +96,15 @@ if $shutdown; then
   exit
 fi
 
+if $active_search_management; then  
+  # need to start MySQL and verify it is up and running before starting SMUI.
+  #log_major "Starting MySQL first to ensure fully available prior to starting SMUI."
+  docker compose up -d mysql  
+  ./mysql/wait-for-mysql.sh
+  services="${services} smui"
+
+fi
+
 docker-compose up -d --build ${services}
 
 echo -e "${MAJOR}Waiting for Solr cluster to start up and all three nodes to be online.${RESET}"
@@ -123,7 +127,7 @@ docker exec solr1 solr zk cp /security.json zk:security.json -z zoo1:2181
 log_minor "waiting for security.json to be available to all Solr nodes"
 ./solr/wait-for-zk-200.sh
 
-echo -e "${MAJOR}Packaging ecommerce configset.${RESET}"
+log_major "Packaging ecommerce configset."
 (cd solr/configsets/ecommerce/conf && zip -r - *) > ./solr/configsets/ecommerce.zip
 log_minor "posting ecommerce.zip configset"
 curl  --user solr:SolrRocks -X PUT --header "Content-Type:application/octet-stream" --data-binary @./solr/configsets/ecommerce.zip "http://localhost:8983/api/cluster/configs/ecommerce"
@@ -209,7 +213,7 @@ if $vector_search; then
 fi
 
 
-echo -e "${MAJOR}Defining relevancy algorithms using ParamSets.${RESET}"
+log_major "Defining relevancy algorithms using ParamSets."
 curl --user solr:SolrRocks -X POST http://localhost:8983/solr/ecommerce/config/params -H 'Content-type:application/json'  -d '{
   "set": {
     "visible_products":{
@@ -385,7 +389,7 @@ fi
 
 
 if $active_search_management; then
-  echo -e "${MAJOR}Setting up SMUI${RESET}"
+  log_major "Setting up SMUI"
   SOLR_INDEX_ID=`curl -S -X PUT -H "Content-Type: application/json" -d '{"name":"ecommerce", "description":"Chorus Webshop"}' http://localhost:9000/api/v1/solr-index | jq -r .returnId`
   curl -S -X PUT -H "Content-Type: application/json" -d '{"name":"product_type"}' http://localhost:9000/api/v1/${SOLR_INDEX_ID}/suggested-solr-field
   curl -S -X PUT -H "Content-Type: application/json" -d '{"name":"title"}' http://localhost:9000/api/v1/${SOLR_INDEX_ID}/suggested-solr-field
@@ -396,27 +400,23 @@ if $offline_lab; then
   log_major "Setting up Quepid"
   ./mysql/wait-for-mysql.sh
 
-  docker-compose run --rm quepid bin/rake db:setup
-  docker-compose run quepid thor user:create -a admin@choruselectronics.com "Chorus Admin" password
+  docker-compose run --rm quepid bundle exec bin/rake db:setup
+  docker-compose run quepid bundle exec thor user:create -a admin@choruselectronics.com "Chorus Admin" password
   log_minor "Setting up Chorus Baseline Relevance case"
   if $local_deploy; then
     solr_collection_url=http://localhost:8983/solr/ecommerce/select
   else
     solr_collection_url=http://chorus.dev.o19s.com:8983/solr/ecommerce/select
   fi
-  docker-compose run quepid thor case:create "Chorus Baseline Relevance" solr ${solr_collection_url} JSONP "id:id, title:title, thumb:img_500x500, name, brand, product_type" "q=#\$query##&useParams=visible_products,querqy_algo" nDCG@10 admin@choruselectronics.com
+  docker-compose run quepid bundle exec thor case:create "Chorus Baseline Relevance" solr ${solr_collection_url} JSONP "id:id, title:title, thumb:img_500x500, name, brand, product_type" "q=#\$query##&useParams=visible_products,querqy_algo" nDCG@10 admin@choruselectronics.com
 
   docker cp ./katas/Broad_Query_Set_rated.csv quepid:/srv/app/Broad_Query_Set_rated.csv
-  docker exec quepid thor ratings:import 1 /srv/app/Broad_Query_Set_rated.csv >> /dev/null
-
+  docker exec quepid bundle exec thor ratings:import 1 /srv/app/Broad_Query_Set_rated.csv >> /dev/null
 
-  log_major "Setting up RRE"
-  docker-compose run rre mvn rre:evaluate
-  docker-compose run rre mvn rre-report:report
 fi
 
 if $observability; then
-  echo -e "${MAJOR}Setting up Grafana${RESET}"
+  log_major "Setting up Grafana"
   curl -u admin:password -S -X POST -H "Content-Type: application/json" -d '{"email":"admin@choruselectronics.com", "name":"Chorus Admin", "role":"admin", "login":"admin@choruselectronics.com", "password":"password", "theme":"light"}' http://localhost:9091/api/admin/users
   curl -u admin:password -S -X PUT -H "Content-Type: application/json" -d '{"isGrafanaAdmin": true}' http://localhost:9091/api/admin/users/2/permissions
   curl -u admin:password -S -X POST -H "Content-Type: application/json" http://localhost:9091/api/users/2/using/1