The tables must be created in the Hive metastore with the alluxio://
location prefix
CREATE TABLE u_user ( userid INT, age INT, gender CHAR(1), occupation STRING, zipcode STRING) ROW FORMAT DELIMITED FIELDS TERMINATED BY '|' LOCATION 'alluxio://bigdata101:19998/ml-100k';
append the Alluxio configuration directory (${ALLUXIO_HOME}/conf
) to the Presto JVM classpath
[root@bigdata101 etc]# cat jvm.config ... -Xbootclasspath/a:/opt/alluxio-2.1.0/conf
把 alluxio-2.1.0-client.jar 复制到 presto 的 plugin/hive-hadoop2/
目录下
[root@bigdata101 hive-hadoop2]# pwd /opt/presto-0.196/plugin/hive-hadoop2 [root@bigdata101 hive-hadoop2]# cp /opt/alluxio-2.1.0/client/alluxio-2.1.0-client.jar .
如果出现 No FileSystem for scheme: alluxio
,说明 jar 包没复制。
把上面修改的两处同步到其他两个节点
重启,测试
[root@bigdata101 presto-0.196]# prestocli --server bigdata101:8881 --catalog hive --schema default presto:default> select * from u_user limit 5; userid | age | gender | occupation | zipcode --------+-----+--------+------------+--------- 1 | 24 | M | technician | 85711 2 | 53 | F | other | 94043 3 | 23 | M | writer | 32067 4 | 24 | M | technician | 43537 5 | 33 | F | other | 15213 (5 rows) Query 20211210_151310_00006_j78ku, FINISHED, 2 nodes Splits: 18 total, 18 done (100.00%) 0:13 [943 rows, 22.1KB] [71 rows/s, 1.68KB/s]
如果出现 failed: No worker nodes available
,在 config.properties
增加 node-scheduler.include-coordinator=true