[SERVER-43990] A command to confirm that all chunks have been moved to the right zone after the initial split Created: 14/Oct/19  Updated: 29/Oct/23  Resolved: 25/Nov/19

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 4.3.3

Type: Task Priority: Major - P3
Reporter: Alyson Cabral (Inactive) Assignee: Marcos José Grillo Ramirez
Resolution: Fixed Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Documented
Backwards Compatibility: Fully Compatible
Sprint: Sharding 2019-11-18, Sharding 2019-12-02
Participants:

 Description   

There is a grey area between the chunks be created and when they move and are owned by the appropriate zone. Users would like to confirm that the chunks are placed according to the zone definition before inserting data. 

I recognize that this happens very quickly in the new initial split model, but this question still comes up fairly frequently. I'd like to provide an easy way for atlas/users to check that it's safe to begin data insertion. 



 Comments   
Comment by Marcos José Grillo Ramirez [ 25/Nov/19 ]

The changes on the output can be tracked with the following ticket:

https://jira.mongodb.org/browse/SERVER-44818

Comment by Githook User [ 21/Nov/19 ]

Author:

{'email': 'marcos.grillo@10gen.com', 'name': 'Marcos José Grillo Ramírez'}

Message: SERVER-43990 A command to confirm that all chunks have been moved to the right zone after the initial split
Branch: master
https://github.com/mongodb/mongo/commit/f48da7a0f83762d214128799923e4bcede800dbe

Comment by Sheeri Cabral (Inactive) [ 28/Oct/19 ]

1. The former is proper in the scope of this ticket. I like the idea of the latter, but if it takes a long time it should be moved to its own ticket. Because the scope of this ticket is to make sure checks are in the right zone. However, I do think it is useful to verify that shards are balanced properly.

2. A cached view is good enough. The systems are constantly changing, so it doesn’t matter if it’s now, or cached from five minutes or an hour ago.

Comment by Kaloian Manassiev [ 21/Oct/19 ]

We can easily add such command, but I would like clarify a few things around its behaviour:

  1. Do you want it to verify only for compliance with zones (i.e., no chunk needs to move because it's in the wrong zone) or is there also merit to expose check for chunks, which need to be moved because there's a draining shard or chunk imbalance? The latter specifically would allow answering whether all hashed sharding chunks been distributed correctly before starting a bulk load?
  2. How important would it be to synchronise this command with concurrent changes to the zone configuration? It might be easier for such a command to use a cached view of the zones rather than take a lock a prevent zones from changing while it's running.
Generated at Thu Feb 08 05:04:41 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.