improvements to KWS deployment. Fixes dmwm#88

das/deploy

@@ -46,18 +46,27 @@ deploy_das_post()
 
   # initialize keylearning db
   rm -f $mongodb_stage/{clean,update}_keylearning_db.js  
+  rm -f $mongodb_stage/keylearning_db-schema-stamp
+  rm -f $mongodb_stage/inputvals_*-schema-stamp
   (echo 'keylearning = db.getSisterDB("keylearning");'
-   echo 'keylearning.dropDatabase();'
-   echo 'inputvals = db.getSisterDB("inputvals");'
+   echo 'keylearning.dropDatabase();') > \
+   $mongodb_stage/clean_keylearning_db.js
+  # it's better to clean DBs separately, datatype_name is the first col to be updated so it will clean...
+  (echo 'inputvals = db.getSisterDB("inputvals");'
    echo 'inputvals.dropDatabase();') > \
-   $mongodb_stage/clean_keylearning_db.js  
+   $mongodb_stage/clean_inputvals_datatype_name.js
+  # clean each collection (currently update_db require clean to be present)
+  for col in group_name primary_dataset_name release_name site_name status_name tier_name
+  do
+  (echo 'inputvals = db.getSisterDB("inputvals");'
+   echo "inputvals.$col.drop();") > \
+   $mongodb_stage/clean_inputvals_$col.js
+  done 
 
   cp $das_root/kws_data/db_dumps/*.js $mongodb_stage/
 
   # initialize whoosh IR index for keyword search
   rm -Rf $root/state/das/kws_index
   mkdir -p $root/state/das/kws_index
   cp $das_root/kws_data/kws_index/* $root/state/das/kws_index/
-
-
 }

mongodb/manage

@@ -48,12 +48,15 @@ cd $STATEDIR
 clean_db()
 {
     set -e
-    if [ -f $STATEDIR/db/${1%_db}.ns ]; then
+    # db is the first item before underscore
+    db=$(echo $1 | cut -f1 -d_)
+    if [ -f $STATEDIR/db/${db}.ns ]; then
         echo "Clean $1"
         mongo --port 8230 stagingarea/clean_$1.js
     fi
     set +e
 }
+
 update_db()
 {
   stamp=$(cat stagingarea/update_$1.js | md5sum)
@@ -106,15 +109,15 @@ start()
   if $started; then
     echo "$ME is ready"
     # keylearning is prerequisite for KWS to function
-    clean_db "keylearning_db"
     update_db "keylearning_db"
     # inputvals is less important, but currently it's not updated live...
     for col in  datatype_name group_name primary_dataset_name release_name site_name status_name tier_name
     do
-      clean_db "inputvals_$col"
       update_db "inputvals_$col" 
     done
     # das
+    # TODO: clean db shall NEVER be called in here, as it do not check for stamps!
+    # TODO: check if this also could potentially cause some strange behaviour in das?!
     clean_db "das_db"
     update_db "mapping_db"