Bundle Cassandra 5.0.9 instead of 4.0.6 in the janusgraph-full distribution

docs/changelog.md

@@ -83,7 +83,7 @@ compile "org.janusgraph:janusgraph-core:1.2.0"
 
 **Installed versions in the Pre-Packaged Distribution:**
 
-* Cassandra 4.0.6
+* Cassandra 5.0.9
 * Elasticsearch 7.14.0
 
 #### Changes
@@ -112,9 +112,9 @@ which itself requires Java 11.
 JanusGraph is now built and tested with Java 11, 17, 21 and 25. The exceptions are the OLAP modules
 (`janusgraph-hadoop` and the Hadoop/Spark based graph computer of the storage backends): Apache Spark
 3.3.x, which TinkerPop's `spark-gremlin` is based on, only runs on Java 8 through 17, so OLAP jobs are
-only supported on Java 11 and 17. The pre-packaged distribution bundles Cassandra 4.0.6, which itself
-only runs on Java 8 and 11, so the embedded Cassandra of the `janusgraph-full` distribution requires
-Java 11 while an external Cassandra cluster can be used from JanusGraph running on any supported Java
+only supported on Java 11 and 17. The pre-packaged distribution bundles Cassandra 5.0.9, which itself
+only runs on Java 11 and 17, so the embedded Cassandra of the `janusgraph-full` distribution requires
+Java 11 or 17 while an external Cassandra cluster can be used from JanusGraph running on any supported Java
 version.
 
 Java 17 and newer enforce strong encapsulation of the JDK internals. JanusGraph itself does not need
@@ -186,8 +186,13 @@ Please review the TinkerPop upgrade documentation for
 ##### Apache Cassandra 5.0 support
 
 Starting from version 1.2.0 JanusGraph supports Apache Cassandra 5.0 as a storage backend.
-The pre-packaged distribution continues to bundle Cassandra 4.0.6; connect JanusGraph to an
-externally managed Cassandra 5.0 cluster to use the new backend.
+The pre-packaged `janusgraph-full` distribution now bundles Cassandra 5.0.9 instead of Cassandra 4.0.6.
+The embedded Cassandra runs on Java 11 and 17 (Cassandra 4.0 only ran on Java 8 and 11). Its
+`cassandra/conf/cassandra.yaml` is regenerated from the stock Cassandra 5.0 configuration with the
+JanusGraph settings (cluster name, `db/cassandra` data directories) applied; `num_tokens` stays at 256 so
+data directories created by earlier `janusgraph-full` distributions keep working. Review the Cassandra
+[5.0 upgrade notes](https://github.com/apache/cassandra/blob/cassandra-5.0/NEWS.txt) before reusing an
+existing `db/cassandra` directory with the new distribution.
 
 ##### ElasticSearch 9 support
 

janusgraph-dist/src/assembly/static/cassandra/bin/cassandra

@@ -12,6 +12,10 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+#
+# This file is derived from the stock bin/cassandra launcher of Apache Cassandra 5.0.9. The only
+# JanusGraph change is the default CASSANDRA_LOG_DIR, which points to the logs directory of the
+# JanusGraph distribution (see the block above the cassandra-env.sh include).
 
 
 # OPTIONS:
@@ -109,23 +113,21 @@ if [ -z "$CASSANDRA_CONF" -o -z "$CLASSPATH" ]; then
     exit 1
 fi
 
+# JanusGraph change: log into the logs directory of the JanusGraph distribution (next to janusgraph.log)
+# instead of cassandra/logs. This has to be set before cassandra-env.sh is sourced because that script
+# applies its own default of $CASSANDRA_HOME/logs.
+if [ -z "$CASSANDRA_LOG_DIR" ]; then
+  CASSANDRA_LOG_DIR=$CASSANDRA_HOME/../logs
+fi
+
 if [ -f "$CASSANDRA_CONF/cassandra-env.sh" ]; then
     . "$CASSANDRA_CONF/cassandra-env.sh"
 fi
 
 if [ -z "$CASSANDRA_LOG_DIR" ]; then
-  # JanusGraph change: All logs in single directory are collected in single directory
-  CASSANDRA_LOG_DIR=$CASSANDRA_HOME/../logs
+  CASSANDRA_LOG_DIR=$CASSANDRA_HOME/logs
 fi
 
-# Special-case path variables.
-case "`uname`" in
-    CYGWIN*|MINGW*) 
-        CLASSPATH=`cygpath -p -w "$CLASSPATH"`
-        CASSANDRA_CONF=`cygpath -p -w "$CASSANDRA_CONF"`
-    ;;
-esac
-
 # Cassandra uses an installed jemalloc via LD_PRELOAD / DYLD_INSERT_LIBRARIES by default to improve off-heap
 # memory allocation performance. The following code searches for an installed libjemalloc.dylib/.so/.1.so using
 # Linux and OS-X specific approaches.
@@ -139,21 +141,27 @@ find_library()
     pattern=$1
     path=$(echo ${2} | tr ":" " ")
 
-    find $path -regex "$pattern" -print 2>/dev/null | head -n 1
+    find $path -regex "$pattern" -print 2>/dev/null | sort -ru | head -n 1
 }
 case "`uname -s`" in
     Linux)
         if [ -z $CASSANDRA_LIBJEMALLOC ] ; then
+            dirs="/lib64 /lib /usr/lib64 /usr/lib "
             which ldconfig > /dev/null 2>&1
             if [ $? = 0 ] ; then
-                # e.g. for CentOS
-                dirs="/lib64 /lib /usr/lib64 /usr/lib `ldconfig -v 2>/dev/null | grep -v '^\s' | sed 's/^\([^:]*\):.*$/\1/'`"
-            else
-                # e.g. for Debian, OpenSUSE
-                dirs="/lib64 /lib /usr/lib64 /usr/lib `cat /etc/ld.so.conf /etc/ld.so.conf.d/*.conf | grep '^/'`"
+                dirs="$dirs `ldconfig -v 2>/dev/null | grep -v '^\s' | sed 's/^\([^:]*\):.*$/\1/'`"
+            fi
+            if [ -f /etc/ld.so.conf ]; then
+                dirs="$dirs `cat /etc/ld.so.conf | grep '^/'`"
+            fi
+            additional_configs=$(ls /etc/ld.so.conf.d/*.conf 2>&1)
+            if [ $? = 0 ]; then
+                for conf in $additional_configs; do
+                    dirs="$dirs `cat $conf | grep '^/'`"
+                done
             fi
             dirs=`echo $dirs | tr " " ":"`
-            CASSANDRA_LIBJEMALLOC=$(find_library '.*/libjemalloc\.so\(\.1\)*' $dirs)
+            CASSANDRA_LIBJEMALLOC=$(find_library '.*/libjemalloc\.so.*' $dirs)
         fi
         if [ ! -z $CASSANDRA_LIBJEMALLOC ] ; then
             export JVM_OPTS="$JVM_OPTS -Dcassandra.libjemalloc=$CASSANDRA_LIBJEMALLOC"