[
https://issues.apache.org/jira/browse/YARN-2578?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14143951#comment-14143951
]
Vinod Kumar Vavilapalli commented on YARN-2578:
-----------------------------------------------
bq. The NM does not fail over correctly when the network cable of the RM is
unplugged or the failure is simulated by a "service network stop" or a firewall
that drops all traffic on the node. The RM fails over to the standby node when
the failure is detected as expected.
I am surprised that RM itself fails over (in the context of firewall rule that
drops traffic) - we never implemented health monitoring like in ZKFC with HDFS.
It seems like if the rpc port gets blocked the RM will not failover as the
embedded ZK continues to use the local loop-back and so doesn't detect the
network failure.
> NM does not failover timely if RM node network connection fails
> ---------------------------------------------------------------
>
> Key: YARN-2578
> URL: https://issues.apache.org/jira/browse/YARN-2578
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Affects Versions: 2.5.1
> Reporter: Wilfred Spiegelenburg
> Attachments: YARN-2578.patch
>
>
> The NM does not fail over correctly when the network cable of the RM is
> unplugged or the failure is simulated by a "service network stop" or a
> firewall that drops all traffic on the node. The RM fails over to the standby
> node when the failure is detected as expected. The NM should than re-register
> with the new active RM. This re-register takes a long time (15 minutes or
> more). Until then the cluster has no nodes for processing and applications
> are stuck.
> Reproduction test case which can be used in any environment:
> - create a cluster with 3 nodes
> node 1: ZK, NN, JN, ZKFC, DN, RM, NM
> node 2: ZK, NN, JN, ZKFC, DN, RM, NM
> node 3: ZK, JN, DN, NM
> - start all services make sure they are in good health
> - kill the network connection of the RM that is active using one of the
> network kills from above
> - observe the NN and RM failover
> - the DN's fail over to the new active NN
> - the NM does not recover for a long time
> - the logs show a long delay and traces show no change at all
> The stack traces of the NM all show the same set of threads. The main thread
> which should be used in the re-register is the "Node Status Updater" This
> thread is stuck in:
> {code}
> "Node Status Updater" prio=10 tid=0x00007f5a6cc99800 nid=0x18d0 in
> Object.wait() [0x00007f5a51fc1000]
> java.lang.Thread.State: WAITING (on object monitor)
> at java.lang.Object.wait(Native Method)
> - waiting on <0x00000000ed62f488> (a org.apache.hadoop.ipc.Client$Call)
> at java.lang.Object.wait(Object.java:503)
> at org.apache.hadoop.ipc.Client.call(Client.java:1395)
> - locked <0x00000000ed62f488> (a org.apache.hadoop.ipc.Client$Call)
> at org.apache.hadoop.ipc.Client.call(Client.java:1362)
> at
> org.apache.hadoop.ipc.ProtobufRpcEngine$Invoker.invoke(ProtobufRpcEngine.java:206)
> at com.sun.proxy.$Proxy26.nodeHeartbeat(Unknown Source)
> at
> org.apache.hadoop.yarn.server.api.impl.pb.client.ResourceTrackerPBClientImpl.nodeHeartbeat(ResourceTrackerPBClientImpl.java:80)
> {code}
> The client connection which goes through the proxy can be traced back to the
> ResourceTrackerPBClientImpl. The generated proxy does not time out and we
> should be using a version which takes the RPC timeout (from the
> configuration) as a parameter.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
