From 4097e29d50ee740d46d446c05101d0dae8ade1bc Mon Sep 17 00:00:00 2001
From: szznax <szznax@daa5b2f2-a927-0410-8b2d-f5f262fa301a>
Date: Tue, 9 Feb 2010 19:45:23 +0000
Subject: [PATCH] part of fix for HER-1736 long pause at startup reading large
 per-job job.log(s) * CrawlJob.java     return early from scanJobLog() when
 job log > 1GB

---
 .../main/java/org/archive/crawler/framework/CrawlJob.java  | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/engine/src/main/java/org/archive/crawler/framework/CrawlJob.java b/engine/src/main/java/org/archive/crawler/framework/CrawlJob.java
index fe8c334c..3bd08d42 100644
--- a/engine/src/main/java/org/archive/crawler/framework/CrawlJob.java
+++ b/engine/src/main/java/org/archive/crawler/framework/CrawlJob.java
@@ -152,6 +152,13 @@ public class CrawlJob implements Comparable<CrawlJob>, ApplicationListener {
         launchCount = 0; 
         if(!jobLog.exists()) return;
         
+        // protect against scanning "large" job logs, which significantly 
+        // slows the web UI
+        if (jobLog.length() > org.apache.commons.io.FileUtils.ONE_GB) { 
+            launchCount = -1; 
+            return; 
+        } 
+        
         try {
             LineIterator lines = FileUtils.lineIterator(jobLog);
             Pattern launchLine = Pattern.compile("(\\S+) (\\S+) Job launched");