Biao Liu created FLINK-5140:
-------------------------------
Summary: NoResourceAvailableException is thrown while starting a
new job
Key: FLINK-5140
URL: https://issues.apache.org/jira/browse/FLINK-5140
Project: Flink
Issue Type: Bug
Components: Cluster Management
Environment: FLIP-6 feature branch
Reporter: Biao Liu
Priority: Minor
Here is the stack trace:
org.apache.flink.runtime.jobmanager.scheduler.NoResourceAvailableException: not
connected to ResourceManager and no slot available
at
org.apache.flink.runtime.instance.SlotPool.internalAllocateSlot(SlotPool.java:281)
at
org.apache.flink.runtime.instance.SlotPool.allocateSlot(SlotPool.java:256)
...
Currently I have to set RestartStrategy to handle this exception. But in some
test cases, I want to test failure about the cluster. It will make this much
more complicated. After discussing with [~tiemsn], maybe we can fix this
problem by executing executionGraph after registered to resource manager.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
