Author: ab Date: Mon Jul 24 07:41:18 2006 New Revision: 425071 URL: http://svn.apache.org/viewvc?rev=425071&view=rev Log: Expire all finished addresses. When sites request long crawl delays this quickly ties down all threads, and lock expiration heppens rarely and proceeds too slowly to remove all expired entries.
Modified: lucene/nutch/trunk/src/plugin/lib-http/src/java/org/apache/nutch/protocol/http/api/HttpBase.java Modified: lucene/nutch/trunk/src/plugin/lib-http/src/java/org/apache/nutch/protocol/http/api/HttpBase.java URL: http://svn.apache.org/viewvc/lucene/nutch/trunk/src/plugin/lib-http/src/java/org/apache/nutch/protocol/http/api/HttpBase.java?rev=425071&r1=425070&r2=425071&view=diff ============================================================================== --- lucene/nutch/trunk/src/plugin/lib-http/src/java/org/apache/nutch/protocol/http/api/HttpBase.java (original) +++ lucene/nutch/trunk/src/plugin/lib-http/src/java/org/apache/nutch/protocol/http/api/HttpBase.java Mon Jul 24 07:41:18 2006 @@ -315,8 +315,8 @@ host = url.getHost(); if (host == null) throw new HttpException("Unknown host for url: " + url); - host = host.toLowerCase(); } + host = host.toLowerCase(); int delays = 0; while (true) { @@ -383,8 +383,6 @@ if (time <= System.currentTimeMillis()) { BLOCKED_ADDR_TO_TIME.remove(host); BLOCKED_ADDR_QUEUE.removeLast(); - } else { - break; } } }