[SERVER-2241] Chunk migrate could consider number of and state of slaves and fail early Created: 17/Dec/10  Updated: 12/Jul/16  Resolved: 06/Mar/14

Status: Closed
Project: Core Server
Component/s: Replication, Sharding
Affects Version/s: None
Fix Version/s: 2.0.0

Type: Improvement Priority: Minor - P4
Reporter: Alberto Lerner Assignee: Unassigned
Resolution: Done Votes: 1
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Participants:

 Description   

> 2) Should we not even start a migrate unless a majory of slaves is up
> when we start?
> 3) Should we require ALL slaves to be up to replicate?

I think yes, we should not migrate if there's no chance of commit.

And no, we should commit a migrate if quorum is present because we'd write
in that case too. We currently push to what we believe to be the majority
of slaves.



 Comments   
Comment by Greg Studer [ 06/Mar/14 ]

This is currently what we do - the moveChunk command can't be run unless a majority of slaves is up, and we don't commit until a majority is caught up.

Generated at Thu Feb 08 02:59:22 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.