[SERVER-11059] Elections can be delayed by some locks Created: 07/Oct/13 Updated: 10/Dec/14 Resolved: 19/Jan/14 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Replication |
| Affects Version/s: | 2.4.6, 2.5.2 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Alexander Komyagin | Assignee: | Matt Dannenberg |
| Resolution: | Duplicate | Votes: | 0 |
| Labels: | elections | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
2-node replica set |
||
| Attachments: |
|
||||||||||||||||
| Issue Links: |
|
||||||||||||||||
| Operating System: | ALL | ||||||||||||||||
| Participants: | |||||||||||||||||
| Description |
|
When the node is doing initial sync and it's in the index build phase, it will not respond to the replSetElect command. Steps to reproduce:
Expected result: to be able to elect a primary if the secondary is doing an initial sync Rationale: in more complicated scenarios involving multiple servers doing initial sync and network splits, this bug can lead to significant (depending on the actual indexes) delays in elections |
| Comments |
| Comment by Matt Dannenberg [ 15/Jan/14 ] |
|
The repro script failed on 2.5.2. But the repro script passed and the repro described in the ticket didn't seem to exhibit the bad behavior at HEAD of master. One of the commits between this ticket being opened and now (between 2.5.4 or and 2.5.5) has fixed this behavior. I believe it may have been this one. |
| Comment by Alexander Komyagin [ 14/Oct/13 ] |
|
Attaching jstest |