Github user mikewalch commented on a diff in the pull request:
https://github.com/apache/accumulo/pull/223#discussion_r103733316
--- Diff: assemble/conf/templates/accumulo-env.sh ---
@@ -53,20 +58,37 @@ shell) JAVA_OPTS=("${JAVA_OPTS[@]}"
${shellHigh_shellLow}) ;;
*) JAVA_OPTS=("${JAVA_OPTS[@]}" ${otherHigh_otherLow}) ;;
esac
-## JVM options set for logging. Review logj4 properties files to see how
they are used.
-JAVA_OPTS=("${JAVA_OPTS[@]}"
+## JVM options set for logging. Review logj4 properties files to see how
they are used.
+JAVA_OPTS=("${JAVA_OPTS[@]}"
"-Daccumulo.log.dir=${ACCUMULO_LOG_DIR}"
-
"-Daccumulo.service.id=${ACCUMULO_CMD}${ACCUMULO_SERVICE_INSTANCE}_$(hostname)"
- "-Daccumulo.audit.log=$(hostname).audit")
+
"-Daccumulo.application=${ACCUMULO_CMD}${ACCUMULO_SERVICE_INSTANCE}_$(hostname)")
case "$ACCUMULO_CMD" in
-monitor) JAVA_OPTS=("${JAVA_OPTS[@]}"
"-Dlog4j.configuration=file:${ACCUMULO_CONF_DIR}/log4j-monitor.properties") ;;
-gc|master|tserver|tracer) JAVA_OPTS=("${JAVA_OPTS[@]}"
"-Dlog4j.configuration=file:${ACCUMULO_CONF_DIR}/log4j-service.properties") ;;
-*) JAVA_OPTS=("${JAVA_OPTS[@]}"
"-Dlog4j.configuration=file:${ACCUMULO_CONF_DIR}/log4j.properties") ;;
+ monitor)
+ JAVA_OPTS=("${JAVA_OPTS[@]}"
"-Dlog4j.configuration=log4j-monitor.properties")
+ ;;
+ gc|master|tserver|tracer)
+ JAVA_OPTS=("${JAVA_OPTS[@]}"
"-Dlog4j.configuration=log4j-service.properties")
+ ;;
+ *)
+ # let log4j use its default behavior (log4j.xml, log4j.properties)
+ true
+ ;;
esac
export JAVA_OPTS
+## External class path items for Java system class loader (dependencies
not included with Accumulo)
+CLASSPATH="$(find "$ZOOKEEPER_HOME"/{,lib}
"$HADOOP_PREFIX"/share/hadoop/{common,common/lib,hdfs,mapreduce,yarn} -maxdepth
1 -name '*.jar' \
+ -and -not -name '*slf4j*' \
+ -and -not -name '*fatjar*' \
+ -and -not -name '*-javadoc*' \
+ -and -not -name '*-sources*.jar' \
+ -and -not -name '*-test*.jar' \
+ -print0 | tr '\0' ':')$CLASSPATH"
--- End diff --
He's doing this because log4j tries to initialize his new appender and will
throw a ClassNotFoundException as the Hadoop & Zookeeper jars set by
`general.classpaths` are loaded later in Java by the Accumulo classloader.
While `general.classpaths` should available to users, it should be empty by
default and the classpath required for Accumulo should be built in
`accumulo-env.sh` before `java` command is called.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
