[
https://issues.apache.org/jira/browse/CASSANDRA-10810?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15390354#comment-15390354
]
Kaide Mu commented on CASSANDRA-10810:
--------------------------------------
Working dtest available:
https://github.com/riptano/cassandra-dtest/compare/master...kdmu:jira-10810
[~pauloricardomg] or [~yukim] could you take a look?
I've tested that it behaves as expected, but there're still two problems:
# After execution, node3.log that is supposed to be saved in logs/last/ is not
present or partial content is lost, could you double-check it?
# You can see that I insert 10k data into node1 and another 10k into node2 with
{{insert_c1c2}} which inserts data with a key from 0 to 10k-1, therefore there
will be always two data with same key. When we use {{query_c1c2}} to check if
data is available on node3 after rebuild, we only can ensure that 10k data is
present since {{query_c1c2}} works with previous keys, the remaining 10k could
be there or not. It's simple to solve, I think we only need to create a new
{{insert_c1c2_from(key)}} and {{query_c1c2_from(key)}} would be enough, but I
don't know if it's necessary.
> Make rebuild operations resumable
> ---------------------------------
>
> Key: CASSANDRA-10810
> URL: https://issues.apache.org/jira/browse/CASSANDRA-10810
> Project: Cassandra
> Issue Type: Wish
> Components: Streaming and Messaging
> Reporter: Jeremy Hanna
> Assignee: Kaide Mu
> Fix For: 3.x
>
>
> Related to CASSANDRA-8942, now that we can resume bootstrap operations, this
> could also be possible with rebuild operations, such as when you bootstrap
> new nodes in a completely new datacenter in two steps.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
