fixed bugs with open files and caching

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@175 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/IndexControl_p.java

@@ -126,6 +126,7 @@ public class IndexControl_p {
                     int i = 0;
                     urlx = new String[index.size()];
                     while (en.hasMoreElements()) urlx[i++] = ((plasmaWordIndexEntry) en.nextElement()).getUrlHash();
+		    index.close();
                 } catch (IOException e) {
                     urlx = new String[0];
                 }
@@ -199,6 +200,7 @@ public class IndexControl_p {
             indexes[0] = switchboard.wordIndex.getEntity(keyhash, true);
             result = yacyClient.transferIndex(yacyCore.seedDB.getConnected(post.get("hostHash", "")), indexes, switchboard.loadedURL);
             prop.put("result", (result == null) ? ("Successfully transferred " + indexes[0].size() + " words in " + ((System.currentTimeMillis() - starttime) / 1000) + " seconds") : result);
+	    try {indexes[0].close();} catch (IOException e) {}
         }
         
 	if (post.containsKey("keyhashsimilar")) {
@@ -206,7 +208,7 @@ public class IndexControl_p {
                 String result = "Sequential List of Word-Hashes:<br>";
                 String hash;
                 int i = 0;
-                while (hashIt.hasNext()) {
+                while ((hashIt.hasNext()) && (i < 256)) {
                     hash = (String) hashIt.next();
                     result += "<a href=\"/IndexControl_p.html?" + 
                     "keystring=" +
@@ -251,7 +253,7 @@ public class IndexControl_p {
                 String result = "Sequential List of URL-Hashes:<br>";
 		String hash;
 		int i = 0;
-                while (hashIt.hasNext()) {
+                while ((hashIt.hasNext()) && (i < 256)) {
 		    hash = (String) hashIt.next();
                     result += "<a href=\"/IndexControl_p.html?" + 
                     "keystring=" +
@@ -376,6 +378,7 @@ public class IndexControl_p {
                 "<span class=\"small\">for every resolveable and deleted URL reference, delete the same reference at every other word where the reference exists (very extensive, but prevents further unresolved references)</span>" +
                 "</td></tr></table></fieldset></form>";
             }
+	    index.close();
             return result;
         }  catch (IOException e) {
             return "";

source/de/anomic/plasma/plasmaSwitchboard.java

@@ -1422,7 +1422,12 @@ public final class plasmaSwitchboard extends serverAbstractSwitch implements ser
                     ee.printStackTrace();
                     return -1;
                 }
-            }
+            } else {
+		// simply close the indexEntities
+		for (int i = 0; i < indexEntities.length; i++) try {
+		    indexEntities[i].close();
+		} catch (IOException ee) {}
+	    }
             return indexCount;
         } else {
             log.logError("Index distribution failed. Too less peers (" + hc + ") received the index, not deleted locally.");
@@ -1514,17 +1519,18 @@ public final class plasmaSwitchboard extends serverAbstractSwitch implements ser
                 indexEntities[i].close();
             } else {
                 // delete complete file
-                if (!(indexEntities[i].deleteComplete())) {
+                if (indexEntities[i].deleteComplete()) {
+                    indexEntities[i].close();
+		} else {
                     indexEntities[i].close();
                     // have another try...
                     if (!(plasmaWordIndexEntity.wordHash2path(plasmaPath, indexEntities[i].wordHash()).delete())) {
                         success = false;
                         log.logError("Could not delete whole Index for word " + indexEntities[i].wordHash());
                     }
-                } else {
-                    indexEntities[i].close();
                 }
             }
+	    indexEntities[i] = null;
         }
         return success;
     }

source/de/anomic/plasma/plasmaWordIndexCache.java

@@ -56,7 +56,8 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
     private static final String oldSingletonFileName = "indexSingletons0.db";
     private static final String newSingletonFileName = "indexAssortment001.db";
     private static final String indexAssortmentClusterPath = "ACLUSTER";
-    private static final int assortmentLimit = 8;
+    private static final int assortmentLimit = 20;
+    private static final int ramcacheLimit = 60;
     
     
     // class variables
@@ -353,15 +354,15 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
             }
 
 	    // print statistics
-	    //for (int k = 0; k < clusterCandidate.length; k++)
+	    for (int k = 0; k < clusterCandidate.length; k++)
-            //    log.logDebug("FLUSH-LIST " + (k + 1) + ": " + clusterCandidate[k].size() + " entries");
+                log.logDebug("FLUSH-LIST " + (k + 1) + ": " + clusterCandidate[k].size() + " entries");
 
             Map.Entry entry;
             int candidateCounter;
 	    count = 0;
 
             // flush high-scores that accumultated too much
-            for (int cluster = clusterCandidate.length; cluster >= 50; cluster--) {
+            for (int cluster = clusterCandidate.length; cluster >= ramcacheLimit; cluster--) {
                 candidateCounter = 0;
                 i = clusterCandidate[cluster - 1].entrySet().iterator();
                 while (i.hasNext()) {
@@ -384,8 +385,8 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                     entry = (Map.Entry) i.next();
                     key = (String) entry.getValue();
                     createTime = (Long) entry.getKey();
-                    if ((createTime != null) && ((System.currentTimeMillis() - createTime.longValue()) > (cluster * 30000))) {
+                    if ((createTime != null) && ((System.currentTimeMillis() - createTime.longValue()) > (20000 + (cluster * 5000)))) {
-                        //log.logDebug("flushing singleton-key " + key + ", count=" + count + ", cachesize=" + cache.size() + ", singleton-size=" + singletons.size());
+                        //log.logDebug("flushing key " + key + ", count=" + count + ", cachesize=" + cache.size() + ", singleton-size=" + singletons.size());
                         count += java.lang.Math.abs(flushFromMem(key, true));
                         candidateCounter += cluster;
                     }
@@ -395,10 +396,10 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
 
 	    // stop flushing if cache is shrinked enough
 	    // avoid as possible to flush high-scores
-	    if (cache.size() < this.maxWords) return count;
+	    if (cache.size() < this.maxWords - 100) return count;
 
             // flush high-scores
-            for (int cluster = java.lang.Math.min(clusterCandidate.length, 50); cluster > assortmentLimit; cluster--) {
+            for (int cluster = java.lang.Math.min(clusterCandidate.length, ramcacheLimit); cluster > assortmentLimit; cluster--) {
                 candidateCounter = 0;
                 i = clusterCandidate[cluster - 1].entrySet().iterator();
                 while (i.hasNext()) {
@@ -410,7 +411,7 @@ public final class plasmaWordIndexCache implements plasmaWordIndexInterface {
                         candidateCounter += cluster;
                         log.logDebug("flushed high-cluster below limit #" + cluster + ", key=" + key + ", count=" + count + ", cachesize=" + cache.size());
                     }
-                    if (cache.size() < this.maxWords) return count;
+                    if (cache.size() < this.maxWords - 100) return count;
                 }
             }