Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Latest member-list may not be received when FinalizeJoinOperation invocation timeouts #9501

Closed
mdogan opened this issue Dec 20, 2016 · 0 comments

Comments

Projects
None yet
2 participants
@mdogan
Copy link
Member

commented Dec 20, 2016

When FinalizeJoinOperation invocation during join timeouts, master can continue processing new join requests and publish new member lists.
But if a member receives/executes FinalizeJoinOperation late, MemberInfoUpdateOperations received before that are all rejected. In this situation, a member may not learn the latest member list until a periodic member list is published (default 5min) or a new member joins.

For example, assume A is master and a single node cluster;

  • B starts up and sends a join request to A
  • A handles the join request, updates member list as [A, B] and sends FinalizeJoinOperation to B
  • A's FinalizeJoinOperation invocation timeouts, B has not processed FinalizeJoinOperation yet
  • C starts up and sends a join request to A
  • A handles the join request, updates member list as [A, B, C] and sends FinalizeJoinOperation to C and MemberInfoUpdateOperation to B
  • B rejects MemberInfoUpdateOperation since it's not joined yet
  • C executes FinalizeJoinOperation and updates member list as [A, B, C]
  • B executes FinalizeJoinOperation that was sent earlier and updates member list as [A, B]

At that point, B cannot learn latest member list [A, B, C] until a periodic member list is published or a new member joins.

Related to #9486

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.