[SERVER-41037] Stepup should kill all user operations(that encounters prepare conflict) before taking RSTL lock in X. - MongoDB Jira

XML

Word

Printable

JSON

Currently, there is a 3 way deadlock. Assume that we are transitioning from secondary to primary(signalDrainComplete).

Thread A (Read thread) acquires RSTL lock in IX and blocked by a prepared txn due to prepare conflict.
Stepup enqueues RSTL lock in X mode. And blocked behind A.
Prepared txn won’t be able commit until we are able to change our state to primary.

~~Alternatively, thread A might have got blocked on prepare conflict due to conflicting DB/ collection lock. For eg. dbhash cmd.~~

EDIT: Blocked due to conflicting mongoDB locks is not possible as secondary prepared txn yield mongoDB locks.

depends on

SERVER-37574 Force reconfig should kill user operations

SERVER-40700 Deadlock between read prepare conflicts and state transitions

related to

SERVER-41035 Rollback should kill all user operations before taking RSTL lock in X.

SERVER-41216 Rename InterruptedDueToStepDown error code to InterruptedDueToReplStateChange

SERVER-37988 recover locks on step up at the beginning of the state transition rather than at the end

Participants:: Githook User, Judah Schvimer, Pavithra Vetriselvan, Suganthi Mani