[
https://issues.apache.org/jira/browse/HADOOP-3673?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12609699#action_12609699
]
dhruba borthakur commented on HADOOP-3673:
------------------------------------------
I think it is good from an architecture point of view to keep the client
relatively lightweight as much as possible. This will allow us to port the
client to many other languages. So, i would like to persist with the current
design (of making the primary datanode invoke recoverBlock).
What if the RPC.Server can me made to support dynamic handler threads? For
example, when the Datanode creates a RPC Server, it can specify a handler count
of 0. The RPC Server code will interpret a count of 0 to mean that the
application wants to create a new thread to service each and every RPC request.
In this case, the RPC Server code will internally create a single handler
thread for this Server. This special handler thread will invoke
callQueue.take() to retrieve a new incoming call, and then fork off a new
thread to process this call. Care has to be taken to ensure that responses from
calls from the same connection are sequentialized and processed in order.
> Deadlock in Datanode RPC servers
> --------------------------------
>
> Key: HADOOP-3673
> URL: https://issues.apache.org/jira/browse/HADOOP-3673
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Affects Versions: 0.18.0
> Reporter: dhruba borthakur
> Priority: Blocker
>
> There is a deadlock scenario in the way Lease Recovery is triggered using the
> Datanode RPC server via HADOOP-3283.
> Each Datanode has dfs.datanode.handler.count handler threads (default of 3).
> These handler threads are used to support the generation-stamp-dance protocol
> as described in HADOOP-1700.
> Let me try to explain the scenario with an example. Suppose, a cluster has
> two datanodes. Also, let's assume that dfs.datanode.handler.count is set to
> 1. Suppose that there are two clients, each writing to a separate file with a
> replication factor of 2. Let's assume that both clients encounter an IO error
> and triggers the generation-stamp-dance protocol. The first client may invoke
> recoverBlock on the first datanode while the second client may invoke
> recoverBlock on the second datanode. Now, each of the datanode will try to
> make a getBlockMetaDataInfo() to the other datanode. But since each datanode
> has only 1 server handler threads, both threads will block for eternity.
> Deadlock!
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
