I guess this is a design limitation of the LucidWorks solr app package. One thing you'd have to do to avoid losing data on restart is to configure Solr to store its data in HDFS. You might be able to do this by setting the solr_opts parameter in the appConfig:
-Dsolr.directoryFactory=HdfsDirectoryFactory -Dsolr.lock.type=none -Dsolr.hdfs.confdir=/etc/hadoop/conf -Dsolr.hdfs.home=${DEFAULT_DATA_DIR}
You would also need to configure automatic failover when creating the collection:
curl "http://server:port/solr/admin/collections?action=CREATE&name=collection_name&numShards=2&autoAddReplicas=true"