Handler command and concurrent DDL can cause node hanging

Bug #1233353 reported by Seppo Jaakola on 2013-09-30
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MySQL patches by Codership
Status tracked in 5.6
5.5
Low
Seppo Jaakola
5.6
Low
Seppo Jaakola
Percona XtraDB Cluster
Status tracked in 5.6
5.5
Undecided
Unassigned
5.6
Undecided
Unassigned

Bug Description

Table handler session can enter into unresolved MDL lock wait with a replicated DDL statement.

The problem happens when a handler session is aborted due to replication conflict. The wsrep abort handler does not release exclusive MDL lock, which remains effective for following sessions. If the connection issues DDL statement as next step, there can be MDL conflict between this DDL and some other replicated DDL.

To reproduce, start two nodes and run two sessions of randgen handler.yy tests against separate nodes.

Changed in codership-mysql:
status: New → In Progress
importance: Undecided → Low
assignee: nobody → Seppo Jaakola (seppo-jaakola)
milestone: none → 5.5.34-24.9
no longer affects: codership-mysql/5.5
Seppo Jaakola (seppo-jaakola) wrote :
Changed in codership-mysql:
status: In Progress → Fix Committed
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers