databases will be downloaded without testing on availability at the first run;...

databases will be downloaded without testing on availability at the first run; while downloading a lockfile is created to stop other psot instances from downloading at the same time

psot/helpers/download_databases_from_s3bucket.sh

@@ -7,15 +7,27 @@ S3BUCKET='psot-databases'
 # list of database folders, downloaded as tar-archives
 list='ghostx uniprot pfam-31'
 
+function make_datadir_and_sync () {
+  
+  echo "Synchronizing databases because of first run"
+  mkdir -p $DATADIR
+  sync_data "$list"
+}
+
 function test_datadir () {
 
   if ! [[ -e $DATADIR && -d $DATADIR ]]; then
-    mkdir $DATADIR
+  
+    echo "Could not find databases in config given path"
+    mkdir -p $DATADIR
     sync_data "$list"
+    
   elif ! all_folders_in_datadir "test_true"; then
+  
+    echo "Missing databases in config given path"
     sync_data "$(all_folders_in_datadir 'return_missing')"
+    
   fi
-
 }
 
 function all_folders_in_datadir () {
@@ -37,27 +49,32 @@ function all_folders_in_datadir () {
   else
     echo "$new_list"
   fi
-
 }
 
 function sync_data () {
 
-  echo "Could not find databases in config given path"
-  echo "Download databases... "
+  echo "Downloading databases... "
 
   cd $DATADIR
 
-  download_list=$1
-  for ele in $download_list; do
-    wget -q "https://${S3BUCKET}.s3.computational.bio.uni-giessen.de/${ele}.tar.gz"
-    tar -zxf "${ele}.tar.gz"
-    rm "${ele}.tar.gz"
-  done
-
-  ln -s -f 'pfam-31/' 'pfam'
-
-  echo "[$(date)] psot (PID $$) downloaded databases to $DATADIR" >> $LOGFILE
-
+  if [[ -e lockfile ]]; then
+    while [[ -e lockfile ]]; do
+      sleep 0.01
+    done
+  else
+    echo "[$(date)] lockfile created from psot (PID $$)" > lockfile
+    download_list=$1
+    for ele in $download_list; do
+      wget -q "https://${S3BUCKET}.s3.computational.bio.uni-giessen.de/${ele}.tar.gz"
+      tar -zxf "${ele}.tar.gz"
+      rm "${ele}.tar.gz"
+    done
+  
+    ln -s -f 'pfam-31/' 'pfam'
+    
+    echo "[$(date)] psot (PID $$) downloaded databases to $DATADIR" >> $LOGFILE
+    rm lockfile
+  fi
 }
 
 function fail () {
@@ -71,5 +88,5 @@ if [[ -e $LOGFILE ]]; then
   test_datadir
 else
   echo "[$(date)] logfile created from psot (PID $$)" > $LOGFILE
-  test_datadir
+  make_datadir_and_sync
 fi