Storage path must be set by the client or fetched from coordinators. about firestorm HOT 4 CLOSED

@zuoliwei with HDFS mode，the following configuration should be set in both client and shuffle server：
client： spark.rss.base.path hdfs://xxxxx
shuffle server: rss.server.hdfs.base.path hdfs://xxxxxx
with Memory_HDFS type, the performance of HDFS plays a key role to make RSS stable

from firestorm.

@colinmjj I edit the conf file as the following:

server.conf
`rss.rpc.server.port 19999

rss.jetty.http.port 19998

rss.storage.basePath hdfs://192.168.130.129:8020/Shuffle_data

rss.storage.type HDFS

rss.coordinator.quorum 192.168.130.131:19999

rss.server.buffer.capacity 1gb

rss.server.partition.buffer.size 150mb

rss.server.read.buffer.capacity 1gb

rss.server.flush.thread.alive 5

rss.server.flush.threadPool.size 10`

spark-defaults.conf
`spark.shuffle.manager org.apache.spark.shuffle.RssShuffleManager

spark.rss.coordinator.quorum 192.168.130.131:19999

spark.rss.storage.type HDFS

rss.storage.base.path hdfs://192.168.130.129:8020/Shuffle_data`

but it does not work, is there any mistakes？

from firestorm.

please check my comments, in spark client, the name of configuration is spark.rss.base.path
BTW, Firestorm won't work well with HDFS, please try MEMORY_LOCAL_HDFS

from firestorm.

please check my comments, in spark client, the name of configuration is spark.rss.base.path BTW, Firestorm won't work well with HDFS, please try MEMORY_LOCAL_HDFS

get， thanks a lot！

from firestorm.