[
https://issues.apache.org/jira/browse/NIFI-4609?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16253711#comment-16253711
]
marco polo commented on NIFI-4609:
----------------------------------
I can see this design decision lending itself toward lower latency ( usually
SiteToSite ) isn't used by so many agents -- so I'm open to moving to a
different protocol with MiNiFi C++
> RemoteSiteListener creates and starts worker threads without bound.
> --------------------------------------------------------------------
>
> Key: NIFI-4609
> URL: https://issues.apache.org/jira/browse/NIFI-4609
> Project: Apache NiFi
> Issue Type: Bug
> Reporter: marco polo
> Assignee: Mark Payne
>
> MiNiFI C++ will continually attempt to reconnect to a NiFi instance. In
> RemoteSiteListener if something happens to the connection the thread does not
> appear to go away and it is unbounded. So if you had many connections come
> in, thread.start will create n threads without bound. This was experienced on
> an RPI where something happened to the connection and we abandoned it on
> MiNiFi C++ ( but socket hasn't been closed yet). Since we don't have insight
> into what's occurring on the server side we make an attempt to reconnect,
> thus creating a new Thread.
> Might suggest an ExecutorService to coordinate these threads and potentially
> an AfterExecute to perform some accounting for visibility.
> https://github.com/apache/nifi/blob/45f82dc855177702a0c1a0e4c38af334b713c278/nifi-nar-bundles/nifi-framework-bundle/nifi-framework/nifi-site-to-site/src/main/java/org/apache/nifi/remote/SocketRemoteSiteListener.java#L310
> [~jeremy.dyer] can provide additional insight.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
