[
https://issues.apache.org/jira/browse/FLINK-12852?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16879396#comment-16879396
]
zhijiang commented on FLINK-12852:
----------------------------------
Thanks for above detail conclusions [~StephanEwen]!
I agree that the option 4 is the most feasible way ATM. The logic of fixed-size
pool is easy and direct to avoid sharing buffers among tasks compared with
dynamic size as now, so tasks are independent with each other, not relying on
any assumptions. We ever realized the fixed-size local pool to avoid the
buffers redistribution in early Blink version.
I understand the option 1 has the same direction with option 4, both with
fixed-size local buffer. The difference is that option 4 takes the
(2*numChannels + floating) as the fixed size of pool for better performance,
and the option 1 takes the numChannels as the fixed size of pool. The option 1
is compatible with users when upgrading, but it might bring performance
regression. The option 4 could keep the performance but might need user to
adjust the config when upgrade failure.
The other three options are indeed complex to implement and might break/rely on
some assumptions to bring potential risks.
> 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
> Components: Runtime / Network
> Affects Versions: 1.7.2, 1.8.1, 1.9.0
> Reporter: Yun Gao
> Assignee: Yun Gao
> Priority: Blocker
> Labels: pull-request-available
> Fix For: 1.9.0
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> 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 1500 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 can solve this problem, but AFAIK in 1.9 it will not include the
> network memory into the ResourceProfile.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
