[SERVER-3062] Read from the nearest slave Created: 09/May/11 Updated: 12/Jul/16 Resolved: 09/May/12 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Internal Client, Sharding |
| Affects Version/s: | None |
| Fix Version/s: | 2.1.2 |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Kristina Chodorow (Inactive) | Assignee: | Ben Becker |
| Resolution: | Done | Votes: | 14 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||
| Participants: | |||||||||||||||||
| Description |
|
as determined by ping time |
| Comments |
| Comment by Pierre Ynard [ 08/Jun/12 ] |
|
Is the problem related to cursors? We have cases with very little constraints on data consistency so I'm a bit confused about what "out of order" means exactly |
| Comment by Eliot Horowitz (Inactive) [ 05/Jun/12 ] |
|
Only when the client re-connects. |
| Comment by Pierre Ynard [ 05/Jun/12 ] |
|
If I understand correctly, if no local replica is available then a non-local one is selected as a fallback, but if a local replica becomes available after that, the client will keep using the non-local, fallback replica that it already selected. Wouldn't it be better if the client automatically switched to the better slave? Is there a scope to the client's connections (besides the node going offline) where at some point the connection will be closed and the slave selection logic will be run again, choosing the better slave that time? |
| Comment by auto [ 09/May/12 ] |
|
Author: {u'login': u'', u'name': u'Ben Becker', u'email': u'ben.becker@10gen.com'}Message: |
| Comment by Scott Hernandez (Inactive) [ 16/Nov/11 ] |
|
We also need an option to read from closest replica, ind. of status (prim/sec). |