For the simple example we currently have three running jobs on the box. Overtime I have seen on Cluster 1 many "Ghost Ports", where a port, lets say 4040 was able to be used for a long time, now is forever held by a ghost-process.
My cluster's boxes are within their own VLAN, with all ports open between them.
Call
spark-shell
Output
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4040. Attempting port 4041.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4041. Attempting port 4042.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4042. Attempting port 4043.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4043. Attempting port 4044.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4044. Attempting port 4045.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4045. Attempting port 4046.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4046. Attempting port 4047.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4047. Attempting port 4048.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4048. Attempting port 4049.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4049. Attempting port 4050.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4050. Attempting port 4051.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4051. Attempting port 4052.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4052. Attempting port 4053.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4053. Attempting port 4054.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4054. Attempting port 4055.
20/03/06 12:54:21 WARN util.Utils: Service 'SparkUI' could not bind on port 4055. Attempting port 4056.
I have tried
- Reseting the box
- Find the jobs by searching for their PIDs by grepping Spark
- Using the 'yarn application' CLI commands
- Killing the cloudera server and agents on spark master/edge node.
Is there anything I can do to get these ports back?