[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15684826#comment-15684826 ] Jian He commented on YARN-5694: --- bq. (Currently, it's only started with manual failover, which doesn't make

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15684625#comment-15684625 ] Daniel Templeton commented on YARN-5694: Test failure is YARN-5362. I should probably add some

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15684494#comment-15684494 ] Hadoop QA commented on YARN-5694: - | (x) *{color:red}-1 overall{color}* | \\ \\ || Vote || Subsystem ||

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15684210#comment-15684210 ] Daniel Templeton commented on YARN-5694: Fair points. I see two ways forward. Either I make all

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15654697#comment-15654697 ] Jian He commented on YARN-5694: --- bq. If we agree that it's bad to have two RMs accidentally sharing the same

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15654259#comment-15654259 ] Daniel Templeton commented on YARN-5694: Yes, but if the RM isn't in HA mode, the fencing is

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15646047#comment-15646047 ] Jian He commented on YARN-5694: --- One other thing is the normal state-store operation already piggybacks a

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15629870#comment-15629870 ] Daniel Templeton commented on YARN-5694: The concern is that the leader election and state store

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15629737#comment-15629737 ] Jian He commented on YARN-5694: --- sorry, I don't quite get it. why is the active thread needed to run in

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15629027#comment-15629027 ] Daniel Templeton commented on YARN-5694: One of the unit test failures is YARN-5043. The other

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15626411#comment-15626411 ] Hadoop QA commented on YARN-5694: - | (x) *{color:red}-1 overall{color}* | \\ \\ || Vote || Subsystem ||

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15626103#comment-15626103 ] Karthik Kambatla commented on YARN-5694: +1, pending Jenkins. > ZKRMStateStore should always

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15623488#comment-15623488 ] Hadoop QA commented on YARN-5694: - | (x) *{color:red}-1 overall{color}* | \\ \\ || Vote || Subsystem ||

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15623378#comment-15623378 ] Karthik Kambatla commented on YARN-5694: I am not very particular about the method name, but do

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15606382#comment-15606382 ] Daniel Templeton commented on YARN-5694: The branch-2.7 patch is out of date. I stopped updating

[jira] [Commented] (YARN-5694) ZKRMStateStore should always start its verification thread to prevent accidental state store corruption

[ https://issues.apache.org/jira/browse/YARN-5694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15606155#comment-15606155 ] Karthik Kambatla commented on YARN-5694: I have minor comments (nits) on the trunk patch: #