[SERVER-9324] replica set step down should fail if it will result in a rollback situation Created: 11/Apr/13 Updated: 14/Apr/16 Resolved: 12/Feb/15 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Replication |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Critical - P2 |
| Reporter: | Scott Hernandez (Inactive) | Assignee: | Unassigned |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Backwards Compatibility: | Fully Compatible |
| Operating System: | ALL |
| Participants: |
| Description |
|
If doing a stepdown results in a rollback situation (no other replica is caught up) then the primary should not step down. Right now we assume that a secondary within a 10 second window will result in them being caught up in time, but we should be more stringent. |
| Comments |
| Comment by Eric Milkie [ 12/Feb/15 ] |
|
As part of the refactor work on replSetStepDown in 3.0, we now wait until a node is fully caught up before stepping down. The period of time we wait is passed as a parameter by the user. |
| Comment by Eric Milkie [ 21/May/14 ] |
|
This issue will be moot once we start recording ops in the oplog as soon as they are read off the network, rather than when they are applied. |