From c8bdd965ece558b0e317877e2df39472bebeef28 Mon Sep 17 00:00:00 2001 From: orbiter Date: Tue, 30 Sep 2008 11:09:46 +0000 Subject: [PATCH] - larger update time for status page - balancer writes cause of robots.txt in log file for crawl delay - removed log output for forced GC - smaller RAM flush for RWI cache, should cause more usage of cache and faster crawling git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5228 6c8d7289-2bf4-0310-a012-ef5d649a1542 --- htroot/Status.html | 2 +- source/de/anomic/crawler/Balancer.java | 2 +- source/de/anomic/plasma/plasmaWordIndex.java | 2 +- source/de/anomic/server/serverMemory.java | 4 ++-- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/htroot/Status.html b/htroot/Status.html index 1ea5e4d75..c1d623339 100644 --- a/htroot/Status.html +++ b/htroot/Status.html @@ -18,7 +18,7 @@ } function init() { loaderBanner = window.setInterval("loadBanner()", 20000); - window.setInterval("reloadGraph()", 5000); + window.setInterval("reloadGraph()", 8000); } diff --git a/source/de/anomic/crawler/Balancer.java b/source/de/anomic/crawler/Balancer.java index 46f40c825..5fdf873d3 100644 --- a/source/de/anomic/crawler/Balancer.java +++ b/source/de/anomic/crawler/Balancer.java @@ -573,7 +573,7 @@ public class Balancer { // in best case, this should never happen if the balancer works propertly // this is only to protection against the worst case, where the crawler could // behave in a DoS-manner - serverLog.logInfo("BALANCER", "forcing fetch delay of " + sleeptime + " millisecond for " + crawlEntry.url().getHost()); + serverLog.logInfo("BALANCER", "forcing crawl-delay of " + sleeptime + " milliseconds for " + crawlEntry.url().getHost() + ((sleeptime > Math.max(minimumLocalDelta, minimumGlobalDelta)) ? " (caused by robots.txt)" : "")); try {synchronized(this) { this.wait(sleeptime); }} catch (final InterruptedException e) {} } diff --git a/source/de/anomic/plasma/plasmaWordIndex.java b/source/de/anomic/plasma/plasmaWordIndex.java index 2d661a4bd..0b3434c61 100644 --- a/source/de/anomic/plasma/plasmaWordIndex.java +++ b/source/de/anomic/plasma/plasmaWordIndex.java @@ -78,7 +78,7 @@ public final class plasmaWordIndex implements indexRI { // environment constants public static final long wCacheMaxAge = 1000 * 60 * 30; // milliseconds; 30 minutes public static final int wCacheMaxChunk = 800; // maximum number of references for each urlhash - public static final int lowcachedivisor = 1200; + public static final int lowcachedivisor = 1800; public static final int maxCollectionPartition = 7; // should be 7 diff --git a/source/de/anomic/server/serverMemory.java b/source/de/anomic/server/serverMemory.java index 485f1ba66..00cd0592a 100644 --- a/source/de/anomic/server/serverMemory.java +++ b/source/de/anomic/server/serverMemory.java @@ -50,9 +50,9 @@ public class serverMemory { final long elapsed = System.currentTimeMillis() - lastGC; if (elapsed > last) { final long free = free(); - System.out.println("vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv"); + //System.out.println("vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv"); System.gc(); - System.out.println("^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ if you see this many times please report to forum"); + //System.out.println("^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ if you see this many times please report to forum"); lastGC = System.currentTimeMillis(); if (log.isFine()) log.logInfo("[gc] before: " + bytesToString(free) + ", after: " + bytesToString(free()) + ", call: " + info); } else if (log.isFine()) {