THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
...
But I think it should not be restarted globally when the speculation execution failover count reach the max-retry-counts.
Black list of node
Most long tail task are caused by cluster problems, so I must ensure speculative execution runs on different node from origin execution.
I will introduce blacklist module into Flink
Yarn
k8s
Mesos
Manage input and output of each ExecutionVertex
...