Add sitemap.size.max
Overrides http.content.limit and file.content.limit when fetching sitemaps (which can be pretty big).
diff --git a/conf/nutch-default.xml b/conf/nutch-default.xml
index 2ba04d2..97fcbe4 100644
--- a/conf/nutch-default.xml
+++ b/conf/nutch-default.xml
@@ -2671,4 +2671,12 @@
Maximum number of redirects to follow.
</description>
</property>
+
+<property>
+ <name>sitemap.size.max</name>
+ <value>52428800</value>
+ <description>
+ Maximum sitemap size in bytes.
+ </description>
+</property>
</configuration>