Github user lvfangmin commented on the issue:
https://github.com/apache/zookeeper/pull/610
I forgot but actually I think I've considered the children of children case
when I created this diff, it actually won't cause the cversion and pzxid
mismatch issue.
The tricky scenario
Github user lvfangmin commented on the issue:
https://github.com/apache/zookeeper/pull/610
After revisiting the logic, I found it's not possible that the direct
children is being creating and serialized after the parent is serialized, but
the children of children might be added during
Github user lvfangmin commented on the issue:
https://github.com/apache/zookeeper/pull/610
Seems there is a Hudson environment problem on Jenkins, it's complaining
about java.lang.NoClassDefFoundError: hudson.model.Computer
---
Github user nkalmar commented on the issue:
https://github.com/apache/zookeeper/pull/610
Looks like the C test failed, can you please re-run if Jenkins throws the
same error? (just add an empty --amend commit).
---
Github user nkalmar commented on the issue:
https://github.com/apache/zookeeper/pull/610
Oh, nice, thanks @lvfangmin !
I will take a look on both then.
---
Github user lvfangmin commented on the issue:
https://github.com/apache/zookeeper/pull/610
@nkalmar the child node deletion problem is still there, which should be
solved in the other PR I sent out recently:
https://github.com/apache/zookeeper/pull/605.
When revisit that, I