[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user lvfangmin commented on the issue: https://github.com/apache/zookeeper/pull/447 @hanm is the new change looks good to you? ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user lvfangmin commented on the issue: https://github.com/apache/zookeeper/pull/447 Thanks @hanm, I've updated the diff based on your comments and rebased onto latest master code. ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user hanm commented on the issue: https://github.com/apache/zookeeper/pull/447 Logic looks good to me. Summarize the change in one sentence: moving global session commit from pre processor to final processor so a global session will not be applied to zkDB until upgrade finished (global session creation committed to quorum). ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user lvfangmin commented on the issue: https://github.com/apache/zookeeper/pull/447 Thanks @anmolnar for review. @hanm please take a look when you have time. ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user hanm commented on the issue: https://github.com/apache/zookeeper/pull/447 I'll give it a review; I missed the original JIRA. ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user lvfangmin commented on the issue: https://github.com/apache/zookeeper/pull/447 @anmolnar do you have time to revisit the code here? ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user lvfangmin commented on the issue: https://github.com/apache/zookeeper/pull/447 @anmolnar I also find it's hard to track when using overwriting the commit, I'll use stack instead. For the tests, I need to inject some failure in the Quorum being set up, that's why I need the customized way I'm doing in my code, I think I prefer to create a separate test class. ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user anmolnar commented on the issue: https://github.com/apache/zookeeper/pull/447 @lvfangmin Got it. Recently we talked about why multiple commits are better for review over squashing. Now I see why. ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user lvfangmin commented on the issue: https://github.com/apache/zookeeper/pull/447 @anmolnar I changed the code based on your comments and did amend, is this the right process? ---
[GitHub] zookeeper issue #447: [ZOOKEEPER-2926] Fix potential data consistency issue ...
Github user anmolnar commented on the issue: https://github.com/apache/zookeeper/pull/447 @lvfangmin Have you amended your commit? I see a lot of changes since I last reviewed this PR. ---