[CONTROLLER-1159] Clustering : Switching behavior to IsolatedLeader can cause disruption on transaction processing Created: 14/Feb/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: Cannot Reproduce 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: 2717

 Description   

When a Leader does not get responses from a majority of it's peers in the election timeout period the Leader switches to the IsolatedLeader behavior. When this happens transactions which are open against the current leader may timeout.

To reproduce follow the steps below,

1. On a single machine configure the controller as the first member in a 3 node cluster.
2. On the same machine run the data store simulator as member-2 and member-3
3. Once the simulators start getting append entries connect mininet to the controller attach mininet to it with 127 switches
4. Watch the log. Soon you will start seeing timeout exceptions.

 Comments   
Comment by Moiz Raja [ 24/Feb/15 ]

From further testing I determined that AppendEntriesReply were not coming in because the dummy datastore was not responding to appendentries. The timeout issues are not reproducible right now.

Comment by Moiz Raja [ 25/Mar/15 ]

Nothing to fix for now

Generated at Wed Feb 07 19:54:50 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.