[SERVER-35771] why elect lower priority node master? Created: 25/Jun/18  Updated: 14/Aug/18  Resolved: 28/Jun/18

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

Type: Bug Priority: Major - P3
Reporter: wendywong Assignee: Nick Brewer
Resolution: Duplicate Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Operating System: ALL
Participants:

 Description   

mongod:3.2.12

os:centos 6.4

A and B node are in the same room,C node in other room ,A.priority=10,B.priority=5,C.priority=1,on the primary node A:cfg=rs.conf();cfg.members[B.node.id]=priority=15;rs.reconfig(cfg);

A node and B node log:VoteRequester: Got no vote from xx.xx.xx.xx:yyyy because: candidate's data is staler than mine, resp

but node is primary 

node A and node B continue to try in Starting an election for a priority takeover,but failed to be primary 

Q1: why the higher priority node A or node B not elected for master?



 Comments   
Comment by wendywong [ 09/Jul/18 ]

thanks very much

Comment by Nick Brewer [ 28/Jun/18 ]

Hi wendywong

I believe what you're running into is a known issue that was addressed in SERVER-23663. I'd suggest that you upgrade to version 3.4.15, as it contains a fix for this behavior.

If you're still running into the problem after upgrading, let us know and we can investigate further.

Regards,
Nick

Comment by wendywong [ 27/Jun/18 ]

hi Nick Brewer 

i doubt some log below:

[SyncSourceFeedback] SyncSourceFeedback error sending update: network error while attempting to run command 'replSetUpdatePosition' on host 'xx.xx.xx.xx:yyyy'
[SyncSourceFeedback] updateUpstream failed: HostUnreachable: network error while attempting to run command 'replSetUpdatePosition' on host 'xx.xx.xx.xx:yyyy' , will retry

VoteRequester: Got no vote from xxxxx because: candidate's data is staler than mine, resp:{ term: 34, voteGranted: false, reason: "candidate's data is staler than mine", ok: 1.0 }

what  does that   means ? how deal with it ?

 

thanks ahead

wendywong

 

Comment by wendywong [ 27/Jun/18 ]

hi Nick Brewer,the node with "_id" : 0,besides,  with "_id"  1 2 3 4  5  already exist,then i add  the other 6 members in the following method

rs.add({host:"xx.xx.xx.xx:yyyy",priority:0,votes:0,hidden:true,tags:{IDC:"BJ"}})

when the newly added members state are secondary,then i promote the one of new members to be primary by increasing the weight ,and then happened: the member with lower priority was elected for master,the higher priority member did not for the reason that VoteRequester: Got no vote from xx.xx.xx.xx:ttttt because: candidate's data is staler than mine, resp:{ term: 32, voteGranted: false, reason: "candidate's data is staler than mine", ok: 1.0 }

 thanks 

WendyWong

Comment by Nick Brewer [ 26/Jun/18 ]

Hi wendywong,

Was the node with "_id" : 0 previously removed from this replica set, and then re-added?

Nick

Comment by wendywong [ 26/Jun/18 ]

hi Nick Brewer,I have uploaded the log,can you see it?

Comment by Nick Brewer [ 25/Jun/18 ]

Hi wendywong

Thanks for your report. I'd like to see the following information:

  • The mongod logs for each of the replica set members, during the time frame when an election is triggered.
  • The full output of rs.conf().

Since this is a public project, you can upload this information to our secure upload portal if you'd prefer. Information that you share there is only visible to MongoDB employees, and it is automatically removed after a period of time.

Thanks,
Nick

Comment by wendywong [ 25/Jun/18 ]

mongod:3.2.12

os:centos 6.4

A and B node are in the same room,C node in other room ,A.priority=10,B.priority=5,C.priority=1,on the primary node A:cfg=rs.conf();cfg.members[B.node.id].priority=15;rs.reconfig(cfg);

A node and B node log:VoteRequester: Got no vote from xx.xx.xx.xx:yyyy because: candidate's data is staler than mine, resp.

but node C  is primary 

node A and node B continue to try in Starting an election for a priority takeover,but failed to be primary 

Q1: why the higher priority node A or node B not elected for master?

 

Generated at Thu Feb 08 04:40:54 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.