[SERVER-21277] Deadlock triggered by dbEval+writeCmd and replSetInitiate on single mongod Created: 04/Nov/15 Updated: 06/Dec/22 Resolved: 31/May/16 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Querying, Replication |
| Affects Version/s: | 3.2.0-rc1 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Robert Guo (Inactive) | Assignee: | Backlog - Replication Team |
| Resolution: | Done | Votes: | 0 |
| Labels: | fuzzer-blacklist, fuzzer-blocker | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Replication
|
||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||
| Operating System: | ALL | ||||||||
| Participants: | |||||||||
| Description |
|
There's a potential deadlock with the global lock and the replication lock. dbEval grabs the Global lock first: then the replication lock in the writeCmd called from extractWriteConcern: replSetInitiate grabs the replication lock first: then the global lock around here: |
| Comments |
| Comment by Robert Guo (Inactive) [ 31/May/16 ] |
|
Closing this issue as "gone away" as it is no longer possible to call rs.initiate() on a standalone mongod after |
| Comment by Kamran K. [ 31/May/16 ] |
|
judah.schvimer / robert.guo, is this bug still possible after fixing |
| Comment by Eric Milkie [ 04/Nov/15 ] |
|
I put this in Planning Bucket A, since it isn't possible to actually do writes successfully (except to local db) on a replSet node that hasn't been initiated yet. This, coupled with waning use of eval, lowers the priority of this ticket. Please comment if you believe otherwise. |