Dear, I meet a strange issue and I am not sure whether it is a Spark usage limitation or a configuration issue.
I run Spark 1.5.1 in standalone mode. There is only one node in my cluster. All services status are ok. While I visit the Spark web address, the Spark URL is spark://c910f04x12.pok.test.com:7077 and I can successfully submit job and start remote Spark shell with the URL and then I change the host long name to the IP address: spark://10.4.12.1:7077 but failed. And then, I check the master output, it caused by message dropped by akka 16/02/22 10:55:06 ERROR ErrorMonitor: dropping message [class akka.actor.ActorSelectionMessage] for non-local recipient [Actor[akka.tcp:// sparkMaster@10.4.12.1:7077/]] arriving at [akka.tcp:// sparkMaster@10.4.12.1:7077] inbound addresses are [akka.tcp:// sparkmas...@c910f04x12.pok.test.com:7077] And then, I stop the service and set the SPARK_MASTER_IP to 10.4.12.1 in spark-evn.sh and then start the service, I found the spark URL change to spark://10.4.12.1:7077 and then I can successfully submit job with the URL of IP format but if I change to host name, the submission will be failed again. And the log similar just change the host name to IP. So, is it my usage error or a usage limitation in Spark, the URL must be exact same string but not replaced by IP or hostname? BTW, I use the nslookup to test my hostname and IP, the result is right. And I also try to add a long name and IP mapping in the /etc/hosts but there is no any help.
