-
Type: Improvement
-
Resolution: Unresolved
-
Priority: Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Storage Execution
-
Execution Team 2022-12-12, Execution Team 2022-12-26, Execution Team 2023-01-09
The deadlocks described in SERVER-71191 and SERVER-44722 are caused because the following conditions are true, in general
- An operation is unkillable by stepdown
- While holding the RSTL in IX mode, an operation takes an X collection lock. If there are any prepared transactions, this blocks.
- The stepdown thread tries to acquire the X lock, but blocks because of a conflict
In this situation, the operation that isn't interrupted by stepdown does not make progress. We should add an assertion to our lock helpers that prevent unkillable operations from taking X locks while also holding the RSTL.
- is related to
-
SERVER-78662 Deadlock with index build, step down, prepared transaction, and MODE_IS coll lock
- Closed
-
SERVER-73036 Investigate potential deadlock with index builds
- Closed
-
SERVER-72898 Investigate if updateShardIdentityConfigString can use MODE_IX lock
- Closed
-
SERVER-72897 Investigate unorthodox locking in ReplicationCoordinatorExternalState
- Closed
- related to
-
SERVER-44722 3 way deadlock can happen between hybrid index build, prepared transactions and stepdown thread on primary that runs index build via coordinator.
- Closed
-
SERVER-71191 Deadlock between index build setup, prepared transaction, and stepdown
- Closed
-
SERVER-75288 Investigate whether the stepdown killop thread should kill operations that hold the RSTL
- Open