Yun Gao created FLINK-12852:
-------------------------------
Summary: Deadlock occurs when requiring exclusive buffer for RemoteInputChannel
Key: FLINK-12852
URL:
https://issues.apache.org/jira/browse/FLINK-12852 Project: Flink
Issue Type: Bug
Affects Versions: 1.9.0
Reporter: Yun Gao
Assignee: Yun Gao
When running tests with an upstream vertex and downstream vertex, deadlock occurs when submitting the job:
{code:java}
"Sink: Unnamed (3/500)" #136 prio=5 os_prio=0 tid=0x00007f2cca81b000 nid=0x38845 waiting on condition [0x00007f2cbe9fe000]
java.lang.Thread.State: TIMED_WAITING (parking)
at sun.misc.Unsafe.park(Native Method)
- parking to wait for <0x000000073ed6b6f0> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
at java.util.concurrent.locks.LockSupport.parkNanos(LockSupport.java:233)
at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(AbstractQueuedSynchronizer.java:2078)
at java.util.concurrent.ArrayBlockingQueue.poll(ArrayBlockingQueue.java:418)
at org.apache.flink.runtime.io.network.buffer.NetworkBufferPool.requestMemorySegments(NetworkBufferPool.java:180)
at org.apache.flink.runtime.io.network.buffer.NetworkBufferPool.requestMemorySegments(NetworkBufferPool.java:54)
at org.apache.flink.runtime.io.network.partition.consumer.RemoteInputChannel.assignExclusiveSegments(RemoteInputChannel.java:139)
at org.apache.flink.runtime.io.network.partition.consumer.SingleInputGate.assignExclusiveSegments(SingleInputGate.java:312)
- locked <0x000000073fbc81f0> (a java.lang.Object)
at org.apache.flink.runtime.io.network.partition.consumer.SingleInputGate.setup(SingleInputGate.java:220)
at org.apache.flink.runtime.taskmanager.Task.setupPartionsAndGates(Task.java:836)
at org.apache.flink.runtime.taskmanager.Task.run(Task.java:598)
at java.lang.Thread.run(Thread.java:834)
{code}
This is due to the required and max of local buffer pool is not the same and there may be over-allocation, when assignExclusiveSegments there are no available memory.
The detail of the scenarios is as follows: The parallelism of both upstream vertex and downstream vertex are 1000 and 500 respectively. There are 200 TM and each TM has 10696 buffers( in total and has 10 slots. For a TM that runs 9 upstream tasks and 1 downstream task, the 9 upstream tasks start first with local buffer pool \{required = 500, max = 2 * 500 + 8 = 1008}, it produces data quickly and each occupy about 990 buffers. Then the DownStream task starts and try to assigning exclusive buffers for 1500 -9 = 1491 InputChannels. It requires 2981 buffers but only 1786 left. Since not all downstream tasks can start, the job will be blocked finally and no buffer can be released, and the deadlock finally occurred.
I think although increasing the network memory solves the problem, the deadlock may not be acceptable. Fined grained resource management [Flink-12761|
https://issues.apache.org/jira/browse/FLINK-12761] can solve this problem, but AFAIK in 1.9 it will not include the network memory into the ResourceProfile. I think the possible solution currently may be one of
# Make the required and max equal for the local buffer pool.
# Add max retrying for allocating exclusive buffers. When exceeding the maximum retrying times, the task will fail and throw an exception that tells users to increase the network memory.
I think the second one may be better.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
Locked
