我运行了几天的应用程序将重新启动。从日志中，我发现了
之类的消息2018-07-15 15:25:47,015 WARN monitor.ContainersMonitorImpl (ContainersMonitorImpl.java:isProcessTreeOverLimit(327)) - Process tree for container: container_e12_1526888270443_0002_02_000059 has processes older than 1 iteration running over the configured limit. Limit=4294967296, current usage = 4295143424
2018-07-15 15:25:47,015 WARN monitor.ContainersMonitorImpl (ContainersMonitorImpl.java:run(516)) - Container [pid=154495,containerID=container_e12_1526888270443_0002_02_000059] is running beyond physical memory limits. Current usage: 4.0 GB of 4 GB physical memory used; 5.7 GB of 8.4 GB virtual memory used. Killing container.
Dump of the process-tree for container_e12_1526888270443_0002_02_000059 :
|- PID PPID PGRPID SESSID CMD_NAME USER_MODE_TIME(MILLIS) SYSTEM_TIME(MILLIS) VMEM_USAGE(BYTES) RSSMEM_USAGE(PAGES) FULL_CMD_LINE
|- 154495 154492 154495 154495 (bash) 0 0 16379904 716 /bin/bash -c /usr/jdk64/jdk1.8.0_77/bin/java -Xms1252m -Xmx1252m -XX:MaxDirectMemorySize=1820m -Dlog.file=/mnt/dfs/4/hadoop/yarn/log/application_1526888270443_0002/container_e12_1526888270443_0002_02_000059/taskmanager.log -Dlogback.configurationFile=file:./logback.xml -Dlog4j.configuration=file:./log4j.properties org.apache.flink.yarn.YarnTaskManager --configDir . 1> /mnt/dfs/4/hadoop/yarn/log/application_1526888270443_0002/container_e12_1526888270443_0002_02_000059/taskmanager.out 2> /mnt/dfs/4/hadoop/yarn/log/application_1526888270443_0002/container_e12_1526888270443_0002_02_000059/taskmanager.err
|- 154502 154495 154495 154495 (java) 37592403 6588980 6142443520 1047903 /usr/jdk64/jdk1.8.0_77/bin/java -Xms1252m -Xmx1252m -XX:MaxDirectMemorySize=1820m -Dlog.file=/mnt/dfs/4/hadoop/yarn/log/application_1526888270443_0002/container_e12_1526888270443_0002_02_000059/taskmanager.log -Dlogback.configurationFile=file:./logback.xml -Dlog4j.configuration=file:./log4j.properties org.apache.flink.yarn.YarnTaskManager --configDir .
所以我知道发生这种情况的原因是容器使用了比我们设置的更多的内存，但是我想知道在杀死该容器之前是否可以转储某东西，以便我可以检查为什么它要使用大量内存