[SERVER-7544] Support priorities (like rs nodes) to control being balancer Created: 02/Nov/12  Updated: 06/Dec/22  Resolved: 19/Apr/18

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

Type: New Feature Priority: Critical - P2
Reporter: Maxim Fedchyshyn Assignee: [DO NOT USE] Backlog - Sharding Team
Resolution: Done Votes: 0
Labels: balancer, cluster, mongos
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
Assigned Teams:
Sharding
Participants:

 Description   

Hello,
I would like to request new feature in cluster environment - ability to specify new BalancerHost during sh.startBalancer().

For example, I have cluster made of number of replicaSets. Each replicaSet have backup servers inside, with priority 0 in order to do only backup job.

It would be also nice idea to make priority levels also working for electing as BalancerHost in the cluster instead of specifying exact Balancers.

Thank you,
Maxim



 Comments   
Comment by Gregory McKeon (Inactive) [ 19/Apr/18 ]

The balancer now runs on the config server, so this has gone away.

Comment by Maxim Fedchyshyn [ 05/Apr/13 ]

Can we change ticket's status to "Waiting for bug fix"?

Thank you,
Maxim

Comment by Maxim Fedchyshyn [ 02/Nov/12 ]

I would better say, I want priorities for cluster balancing election between mongos working the same way as priorities in replicaSets between mongod.
It would allow to specify which mongos instance would be elected as balancer, which would be elected in case of previous node failure, and which wouldn't be elected at all.

Comment by Scott Hernandez (Inactive) [ 02/Nov/12 ]

Would it be fair to say that you want a mongos instance that won't act as a balancer at all for that use?

Comment by Maxim Fedchyshyn [ 02/Nov/12 ]

No, there is any performance issues.
IT is mostly organizing and structuring issue, where I would like to see hosts doing exact job they were provisioned to and not involved into any other activity.

Comment by Scott Hernandez (Inactive) [ 02/Nov/12 ]

The balancer doesn't actually move the data so there shouldn't be much io, nor cpu to act as the balancer. Is there a performance issue when the balancer runs on the backup node?

Generated at Thu Feb 08 03:14:50 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.