Loading...

XML

Word

Printable

JSON

Type: Bug
Resolution: Fixed
Priority: Major - P3
Fix Version/s: 3.4.16, 3.6.6, 4.0.1, 4.1.1
Affects Version/s: None
Component/s: Replication
Labels:
- initialSync

Backwards Compatibility:
Fully Compatible
Operating System:
ALL
Backport Requested:

v4.0, v3.6, v3.4
Sprint:
Repl 2018-06-18
Linked BF Score:
0
Confidence Status:
None
Work Order:
3

Aha! Reference:
None
Tracking Level:
None
Risk Status:
None
Exec Notes:
None
Goal Name:
None
Goal Link:
None

When called with initialSync:1, replSetGetStatus locks the ReplicationCoordinatorImpl mutex, then the InitialSyncer mutex. The InitialSyncer in the course of normal operation will acquire the same locks in the opposite order. Result: deadlock.

I do not think it is necessary to hold the ReplicationCoordinatorImpl mutex while getting the initialSync status, so the fix may be as simple as getting the initialSync status before acquiring the mutex.

is related to

SERVER-35372 replSetSyncFrom can cause deadlock between ReplicationCoordinator and InitialSyncer

Closed

related to

SERVER-28840 replSetSyncFrom causes InitialSyncer and ReplicationCoordinator to acquire each other's mutexes in opposite orders

Closed

Assignee:: Suganthi Mani
Reporter:: Matthew Russotto
Participants:: Githook User, Matthew Russotto, Suganthi Mani
Votes:: 0 Vote for this issue
Watchers:: 8 Start watching this issue

Created:: Apr 30 2018 09:15:02 PM UTC
Updated:: Oct 29 2023 10:32:13 PM UTC
Resolved:: Jun 13 2018 07:11:42 PM UTC
Confidence Status Last Update:: 11/Jun/18 4:28 PM

Details

Description

Attachments

Issue Links

Activity

People

Dates