[CONTROLLER-1210] Clustering : Reduce sending of duplicate replication messages Created: 16/Mar/15  Updated: 25/Mar/15  Resolved: 25/Mar/15

Status: Resolved
Project: controller
Component/s: mdsal
Affects Version/s: Post-Helium
Fix Version/s: None

Type: Bug
Reporter: Moiz Raja Assignee: Moiz Raja
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All


External issue ID: 2849
Priority: High

 Description   

With Raft not all append entries are expected to reach the followers
so it is a requirement that when acknowledgement of an append entries
is not received we send a duplicate append entries.

In replication scenarios where a lot of rapid replication messages
are being generated the lack of equally rapid acknowledgement results
in a lot of duplicate messages being generated. This later results in
a cycle when the replies for those duplicate replication messages start
arriving and result in even more duplicate replication messages being
sent.

We need to disallow a replication message
to be generated for a follower if the nextIndex for that follower
has not changed and the time since the last replication message
does not exceed the heartbeat interval.



 Comments   
Comment by Moiz Raja [ 17/Mar/15 ]

https://git.opendaylight.org/gerrit/#/c/16723/

Generated at Wed Feb 07 19:54:57 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.