[SERVER-8887] Improve the balancer start and stop commands Created: 07/Mar/13 Updated: 06/Dec/22 Resolved: 31/Mar/16 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Sharding |
| Affects Version/s: | 2.2.3 |
| Fix Version/s: | None |
| Type: | Improvement | Priority: | Minor - P4 |
| Reporter: | David Hows | Assignee: | [DO NOT USE] Backlog - Sharding Team |
| Resolution: | Duplicate | Votes: | 1 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||
| Assigned Teams: |
Sharding
|
||||||||||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||||||||||
| Participants: | |||||||||||||||||
| Description |
|
Currently there are a few implicit assumptions about the success of the update to the settings collection when we run the sh.startBalancer and sh.stopBalancer commands. Right now we assume that an update to config.settings will always succeed. We should remove these implicit assumptions and replace them with checks and retries. |
| Comments |
| Comment by Andy Schwerin [ 31/Mar/16 ] |
|
sh.start/stop balancer methods now return errors if they do not succeed. |
| Comment by Andy Schwerin [ 22/Jul/15 ] |
|
I believe that improved balancer control should fall out of any project to improve the management of metadata life cycles. |