Hi, Igntrs!
While testing transaction failover recovery mechanisms I came across a bug.
When we have multiple consequent transactions changing the same key and
backup node crashes during prepare phase, not all changes are applied.
This bug is flaky and arises only if there are 2 or more consequent
transactions changing the same key.
I created ticket for it [1] and attached reproducer.
Ps primary or originating node crash doesn’t lead to such a bug.
[1] :
https://issues.apache.org/jira/browse/IGNITE-7591--
*Best Regards,*
*Kuznetsov Aleksey*
Locked
