Merge pull request IQSS#8830 from poikilotherm/7000-mpconfig-rserve

7000 mpconfig rserve

Author: kcondon

doc/sphinx-guides/source/installation/config.rst

@@ -1573,27 +1573,60 @@ Can also be set via *MicroProfile Config API* sources, e.g. the environment vari
 dataverse.rserve.host
 +++++++++++++++++++++
 
-Host name for Rserve, used for tasks that require use of R (to ingest RData files and to save tabular data as RData frames).
+Host name for Rserve, used for tasks that require use of R (to ingest RData
+files and to save tabular data as RData frames).
+
+Defaults to ``localhost``.
+
+Can also be set via *MicroProfile Config API* sources, e.g. the environment
+variable ``DATAVERSE_RSERVE_HOST``.
 
 dataverse.rserve.port
 +++++++++++++++++++++
 
-Port number for Rserve, used for tasks that require use of R (to ingest RData files and to save tabular data as RData frames).
+Port number for Rserve, used for tasks that require use of R (to ingest RData
+files and to save tabular data as RData frames).
+
+Defaults to ``6311`` when not configured or no valid integer.
+
+Can also be set via *MicroProfile Config API* sources, e.g. the environment
+variable ``DATAVERSE_RSERVE_PORT``.
 
 dataverse.rserve.user
 +++++++++++++++++++++
 
-Username for Rserve, used for tasks that require use of R (to ingest RData files and to save tabular data as RData frames).
+Username for Rserve, used for tasks that require use of R (to ingest RData
+files and to save tabular data as RData frames).
+
+Defaults to ``rserve``.
+
+Can also be set via *MicroProfile Config API* sources, e.g. the environment
+variable ``DATAVERSE_RSERVE_USER``.
 
 dataverse.rserve.password
 +++++++++++++++++++++++++
 
-Password for Rserve, used for tasks that require use of R (to ingest RData files and to save tabular data as RData frames).
+Password for Rserve, used for tasks that require use of R (to ingest RData
+files and to save tabular data as RData frames).
+
+Defaults to ``rserve``.
+
+Can also be set via *MicroProfile Config API* sources, e.g. the environment
+variable ``DATAVERSE_RSERVE_PASSWORD``.
 
 dataverse.rserve.tempdir
 ++++++++++++++++++++++++
 
-Temporary directory used by Rserve (defaults to /tmp/Rserv). Note that this location is local to the host on which Rserv is running (specified in ``dataverse.rserve.host`` above). When talking to Rserve, Dataverse needs to know this location in order to generate absolute path names of the files on the other end. 
+Temporary directory used by Rserve (defaults to ``/tmp/Rserv``). Note that this
+location is local to the host on which Rserv is running (specified in
+``dataverse.rserve.host`` above). When talking to Rserve, Dataverse needs to
+know this location in order to generate absolute path names of the files on the
+other end.
+
+Defaults to ``/tmp/Rserv``.
+
+Can also be set via *MicroProfile Config API* sources, e.g. the environment
+variable ``DATAVERSE_RSERVE_TEMPDIR``.
 
 .. _dataverse.dropbox.key:
 

src/main/java/edu/harvard/iq/dataverse/ingest/tabulardata/impl/plugins/rdata/RDATAFileReader.java

@@ -31,6 +31,7 @@
 import javax.inject.Inject;
 
 // Rosuda Wrappers and Methods for R-calls to Rserve
+import edu.harvard.iq.dataverse.settings.JvmSettings;
 import org.rosuda.REngine.REXP;
 import org.rosuda.REngine.REXPMismatchException;
 import org.rosuda.REngine.RList;
@@ -88,10 +89,10 @@ public class RDATAFileReader extends TabularDataFileReader {
   static private String RSCRIPT_WRITE_DVN_TABLE = "";
 
   // RServe static variables
-  private static String RSERVE_HOST = System.getProperty("dataverse.rserve.host");
-  private static String RSERVE_USER = System.getProperty("dataverse.rserve.user");
-  private static String RSERVE_PASSWORD = System.getProperty("dataverse.rserve.password");
-  private static int RSERVE_PORT;
+  private final String RSERVE_HOST;
+  private final int RSERVE_PORT;
+  private final String RSERVE_USER;
+  private final String RSERVE_PASSWORD;
 
   // TODO: 
   // we're not using these time/data formats for anything, are we?
@@ -138,24 +139,6 @@ public class RDATAFileReader extends TabularDataFileReader {
    * This is primarily to construct the R-Script
    */
   static {
-    /*
-     * Set defaults fallbacks for class properties
-     */
-    if (RSERVE_HOST == null)
-      RSERVE_HOST = "localhost";
-
-    if (RSERVE_USER == null)
-      RSERVE_USER = "rserve";
-
-    if (RSERVE_PASSWORD == null)
-      RSERVE_PASSWORD = "rserve";
-
-    if (System.getProperty("dataverse.ingest.rserve.port") == null)
-      RSERVE_PORT = 6311;
-    else
-      RSERVE_PORT = Integer.parseInt(System.getProperty("dataverse.rserve.port"));
-
-    
     // Load R Scripts into memory, so that we can run them via R-serve
     RSCRIPT_WRITE_DVN_TABLE = readLocalResource("scripts/write.table.R");
     RSCRIPT_GET_DATASET = readLocalResource("scripts/get.dataset.R");
@@ -451,7 +434,20 @@ public RDATAFileReader(TabularDataFileReaderSpi originator) {
 
     super(originator);
 
-    
+    // These settings have sane defaults in resources/META-INF/microprofile-config.properties,
+    // ready to be overridden by a sysadmin. Every time a file would be read with this file reader,
+    // a new reader will be created, reading from the cached config source settings with minimal overhead.
+    this.RSERVE_HOST = JvmSettings.RSERVE_HOST.lookup();
+    int port;
+    try {
+      port = JvmSettings.RSERVE_PORT.lookup(Integer.class);
+    } catch (IllegalArgumentException e) {
+      LOG.log(Level.SEVERE, "Could not parse value for " + JvmSettings.RSERVE_PORT.getScopedKey() + ", defaulting to 6311", e);
+      port = 6311;
+    }
+    this.RSERVE_PORT = port;
+    this.RSERVE_USER = JvmSettings.RSERVE_USER.lookup();
+    this.RSERVE_PASSWORD = JvmSettings.RSERVE_PASSWORD.lookup();
 
     LOG.fine("RDATAFileReader: INSIDE RDATAFileReader");
 

src/main/java/edu/harvard/iq/dataverse/rserve/RemoteDataFrameService.java

@@ -41,6 +41,7 @@
 import java.util.Set;
 import java.util.logging.Logger;
 
+import edu.harvard.iq.dataverse.settings.JvmSettings;
 import org.apache.commons.io.IOUtils;
 
 import org.apache.commons.lang3.RandomStringUtils;
@@ -72,57 +73,33 @@ public class RemoteDataFrameService {
 
     private static String TMP_TABDATA_FILE_EXT = ".tab";
     private static String TMP_RDATA_FILE_EXT = ".RData";
-
-    private static String RSERVE_HOST = null;
-    private static String RSERVE_USER = null;
-    private static String RSERVE_PWD = null;    
-    private static int    RSERVE_PORT = -1;
+    
+    // These settings have sane defaults in resources/META-INF/microprofile-config.properties,
+    // ready to be overridden by a sysadmin
+    private final String RSERVE_HOST;
+    private final String RSERVE_USER;
+    private final String RSERVE_PWD;
+    private final int    RSERVE_PORT;
+    private final String RSERVE_TMP_DIR;
 
     private static String DATAVERSE_R_FUNCTIONS = "scripts/dataverse_r_functions.R";
     private static String DATAVERSE_R_PREPROCESSING = "scripts/preprocess.R";
-                    
-    public static String LOCAL_TEMP_DIR = System.getProperty("java.io.tmpdir");
-    public static String RSERVE_TMP_DIR=null;
 
     public String PID = null;
     public String tempFileNameIn = null;
     public String tempFileNameOut = null;
- 
-    static {
-    
-        RSERVE_TMP_DIR = System.getProperty("dataverse.rserve.tempdir");
-        
-        if (RSERVE_TMP_DIR == null){
-            RSERVE_TMP_DIR = "/tmp/";            
-        }
-        
-        RSERVE_HOST = System.getProperty("dataverse.rserve.host");
-        if (RSERVE_HOST == null){
-            RSERVE_HOST= "localhost";
-        }
-        
-        RSERVE_USER = System.getProperty("dataverse.rserve.user");
-        if (RSERVE_USER == null){
-            RSERVE_USER= "rserve";
-        }
-        
-        RSERVE_PWD = System.getProperty("dataverse.rserve.password");
-        if (RSERVE_PWD == null){
-            RSERVE_PWD= "rserve";
-        }
-        
-
-        if (System.getProperty("dataverse.rserve.port") == null ){
-            RSERVE_PORT= 6311;
-        } else {
-            RSERVE_PORT = Integer.parseInt(System.getProperty("dataverse.rserve.port"));
-        }
-
-    }
-
-   
 
     public RemoteDataFrameService() {
+        // These settings have sane defaults in resources/META-INF/microprofile-config.properties,
+        // ready to be overridden by a sysadmin. Config sources have their own caches, so adding
+        // these here means the setting can be changed dynamically without too much overhead.
+        this.RSERVE_HOST = JvmSettings.RSERVE_HOST.lookup();
+        this.RSERVE_USER = JvmSettings.RSERVE_USER.lookup();
+        this.RSERVE_PWD = JvmSettings.RSERVE_PASSWORD.lookup();
+        this.RSERVE_PORT = JvmSettings.RSERVE_PORT.lookup(Integer.class);
+        this.RSERVE_TMP_DIR = JvmSettings.RSERVE_TEMPDIR.lookup();
+        
+        
         // initialization
         PID = RandomStringUtils.randomNumeric(6);
 
@@ -703,15 +680,12 @@ public Map<String, String> runDataFrameRequest(RJobRequest jobRequest, RConnecti
     public File transferRemoteFile(RConnection connection, String targetFilename,
             String tmpFilePrefix, String tmpFileExt, int fileSize) {
 
-        // set up a local temp file: 
-        
+        // set up a local temp file:
         File tmpResultFile = null;
-        String resultFile = tmpFilePrefix + PID + "." + tmpFileExt;
-
         RFileInputStream rInStream = null;
         OutputStream outbr = null;
         try {
-            tmpResultFile = new File(LOCAL_TEMP_DIR, resultFile);
+            tmpResultFile = File.createTempFile(tmpFilePrefix + PID, "."+tmpFileExt);
             outbr = new BufferedOutputStream(new FileOutputStream(tmpResultFile));
             // open the input stream
             rInStream = connection.openFile(targetFilename);

src/main/java/edu/harvard/iq/dataverse/settings/JvmSettings.java

@@ -44,6 +44,14 @@ public enum JvmSettings {
     FQDN(PREFIX, "fqdn"),
     SITE_URL(PREFIX, "siteUrl"),
 
+    // RSERVE CONNECTION
+    SCOPE_RSERVE(PREFIX, "rserve"),
+    RSERVE_HOST(SCOPE_RSERVE, "host"),
+    RSERVE_PORT(SCOPE_RSERVE, "port", "dataverse.ingest.rserve.port"),
+    RSERVE_USER(SCOPE_RSERVE, "user"),
+    RSERVE_PASSWORD(SCOPE_RSERVE, "password"),
+    RSERVE_TEMPDIR(SCOPE_RSERVE, "tempdir"),
+    
     // API SETTINGS
     SCOPE_API(PREFIX, "api"),
     API_SIGNING_SECRET(SCOPE_API, "signing-secret"),

src/main/resources/META-INF/microprofile-config.properties

@@ -14,6 +14,13 @@ dataverse.db.port=5432
 dataverse.db.user=dataverse
 dataverse.db.name=dataverse
 
+# RSERVE
+dataverse.rserve.host=localhost
+dataverse.rserve.port=6311
+dataverse.rserve.user=rserve
+dataverse.rserve.password=rserve
+dataverse.rserve.tempdir=/tmp/Rserv
+
 # OAI SERVER
 dataverse.oai.server.maxidentifiers=100
 dataverse.oai.server.maxrecords=10