[ https://issues.apache.org/jira/browse/IGNITE-6442?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16173393#comment-16173393 ]
Andrey Gura edited comment on IGNITE-6442 at 9/20/17 9:34 PM: -------------------------------------------------------------- [~VitaliyB] It's bad idea to check that deadlock detection procedure was initiated using log messages. In all tests we have expectation about deadlock detection: at least one transaction should detect deadlock. It's enough for tests purposes and more reliable because can't be affected by log message changes. Please, fix it. was (Author: agura): [~VitaliyB] It's bad idead to check that deadlock detection procedure was initiated using log messages. In all tests we have expectation about deadlock detection: at least one transaction should detect deadlock. It's enough for tests purposes and more reliable because can't be affected by log message changes. Please, fix it. > Deadlock detection doesn't execute. > ----------------------------------- > > Key: IGNITE-6442 > URL: https://issues.apache.org/jira/browse/IGNITE-6442 > Project: Ignite > Issue Type: Bug > Reporter: Vitaliy Biryukov > Assignee: Vitaliy Biryukov > Fix For: 2.3 > > > In case of a configuration with near cache and if all entities of one of the > transactions involved in the deadlock are on the node being the initiator of > this transaction, then immediately after the timeout, the transaction rolls > back (without calling DD). -- This message was sent by Atlassian JIRA (v6.4.14#64029)