accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Elser (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-4456) Fix exclusion on answering RPCs when lock is not acquired
Date Wed, 22 Feb 2017 19:14:44 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-4456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15879005#comment-15879005
] 

Josh Elser commented on ACCUMULO-4456:
--------------------------------------

bq. I'm not sure I fully understand what the replication service is doing, but it's possibly
different

bq. There's no reason that the replication thrift server should be any different than the
others.

Again, I don't know why it would be different than the others.

bq. There seem to be 3 layers of service wrappers here

{code}
    ReplicationCoordinator.Iface haReplicationProxy = HighlyAvailableServiceWrapper.service(impl,
this);
    ReplicationCoordinator.Processor<ReplicationCoordinator.Iface> replicationCoordinatorProcessor
= new ReplicationCoordinator.Processor<>(RpcWrapper.service(
        impl, new ReplicationCoordinator.Processor<>(haReplicationProxy)));
{code}

This is what you're referring to? Specifically 

{code}
RpcWrapper.service(impl, new ReplicationCoordinator.Processor<>(haReplicationProxy))
{code}

Not being 

{code}
RpcWrapper.service(haReplicationProxy, new ReplicationCoordinator.Processor<>(haReplicationProxy))
{code}

This sounds exactly like the kind of unintentional omission I meant earlier. The change https://github.com/apache/accumulo/pull/219/files#diff-8328b7e7fcdb9b134f586e736090cd0bL1161
makes sense.



> Fix exclusion on answering RPCs when lock is not acquired
> ---------------------------------------------------------
>
>                 Key: ACCUMULO-4456
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-4456
>             Project: Accumulo
>          Issue Type: Sub-task
>          Components: client, master
>            Reporter: Josh Elser
>            Assignee: Josh Elser
>             Fix For: 2.0.0
>
>          Time Spent: 2h 50m
>  Remaining Estimate: 0h
>
> [~kturner] asked me a good question about what the client does when it tries to talk
to a master which has recently lost its active status, how does the client handle the thrown
exception?
> Should run a quick local test.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message