Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 320728 - TCPConnector should serve the channels in a round-robin way
Summary: TCPConnector should serve the channels in a round-robin way
Status: CLOSED FIXED
Alias: None
Product: EMF
Classification: Modeling
Component: cdo.net4j (show other bugs)
Version: 4.0   Edit
Hardware: PC Linux
: P3 normal (vote)
Target Milestone: ---   Edit
Assignee: Eike Stepper CLA
QA Contact: Eike Stepper CLA
URL:
Whiteboard: offline-xx
Keywords:
Depends on:
Blocks:
 
Reported: 2010-07-23 10:45 EDT by Cyril Jaquier CLA
Modified: 2011-06-23 03:42 EDT (History)
0 users

See Also:

Attachments
Patch - for future reference (12.00 KB, patch)
2010-07-24 08:28 EDT, Eike Stepper CLA 		no flags Details | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Cyril Jaquier CLA 2010-07-23 10:45:02 EDT 
Build Identifier: HEAD

Currently, the implementation of TCPConnector uses a LinkedBlockingList to queue the channels (FIFO).

If a channel C1 sends a lot of data, other channels will be able to send their data only after all the ByteBuffers of C1 have been transmitted.

This is a problem when the HeartBeatProtocol is enabled and a cloned repository is synchronizing over a slow network connection. In such case, the heartbeats are not sent on time and the client thinks that the connection is dead.

A round-robin scheduling of the channels solves this issue.

Reproducible: Always
Comment 1 Eike Stepper CLA 2010-07-24 08:28:34 EDT 
Created attachment 175138 [details]
Patch - for future reference

I implemented a RoundRobinBlockingQueue in net4j.util.
Comment 2 Eike Stepper CLA 2010-07-24 08:33:20 EDT 
Committed to HEAD
Comment 3 Eike Stepper CLA 2011-06-23 03:42:08 EDT 
Available in R20110608-1407