Currently, in order to build Scribe with Hadoop support (--enable-hdfs), you need to specify the CPPFLAGS/LDFLAGS manually to build/link with java and hdfs. Eg:
./configure --with-hadooppath=/usr/local/hadoop --enable-hdfs CPPFLAGS="-I/usr/local/java/include -I/usr/local/java/include/linux" LDFLAGS="-ljvm -lhdfs"
It would be nice to add some intelligent autoconf macros to clean this up.