[SERVER-25838] Network split course OplogOutOfOrder error Created: 28/Aug/16  Updated: 15/Nov/21  Resolved: 29/Aug/16

Status: Closed
Project: Core Server
Component/s: Replication
Affects Version/s: 3.2.3
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Zhang Youdong Assignee: Eric Milkie
Resolution: Duplicate Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Attachments: PDF File OplogOutOfOrder problem.pdf    
Issue Links:
Duplicate
duplicates SERVER-19836 Replace priority election logic for p... Closed
Operating System: ALL
Sprint: Repl 2016-08-29
Participants:

 Description   

The problem is described in the attachment, I cannot find the way to reproduce the problem,but it did happened in our online environment.



 Comments   
Comment by Zhang Youdong [ 29/Aug/16 ]

Eric Milkie

Thank you for your reply, I will reconfig our replica set to use protocol version 1 ASAP.

Comment by Eric Milkie [ 29/Aug/16 ]

Protocol version 1 uses a new algorithm for priority takeover, which avoids problems like the one you experienced here. I recommend upgrading to pv1 (which is the default for new replicasets) to avoid having the problem recur.

Comment by Zhang Youdong [ 29/Aug/16 ]

MongoDB 3.2.3 with replication protocol version 0

Comment by Eric Milkie [ 28/Aug/16 ]

Hi Zhang Youdong
In this replicaset, are you using protocol version 0 or protocol version 1? This value appears in the replica set config.

If you are using protocol version 1, I expect that "t10" will not happen (the code logic does not do this), and "t11" is not possible (no other nodes would vote for node2 in its election because it is not freshest).

Generated at Thu Feb 08 04:10:22 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.