[
https://issues.apache.org/jira/browse/CASSANDRA-18804?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17795459#comment-17795459
]
David Capwell commented on CASSANDRA-18804:
-------------------------------------------
recent changes +1ed from [~benedict] .
For context, [~bdeggleston] approved the patch but I saw a BurnTest issue so
reached out to [~benedict] so a few changes were added to make that test stable
1) min epoch selection doesn't handle cases where ranges were moved, causing
the wrong min epoch to be selected; the change was to return the intersected
shards rather than the shards that intersected and are not synced
2) BurnTest is now scoped to hash (0, 1 << 16) to make it easier to debug; this
is a java static variable config so larger ranges may be tested on demand.
3) reads are now blocking rather than async, this should be brought up in a
different Jira
4) BurnTest now fails if we don't learn the final status for each txn; before
we would just log this making it hard to tell this happened
5) CommandTimeseries may NPE due to execute_at being null
6) AbstractRanges.indexOf would return first range found from binary search
rather than the first in the list. This causes different nodes to select
different home keys which leads to failures in SPL
> (Accord): Bug fixes from CASSANDRA-18675 to better support adding keyspaces
> ---------------------------------------------------------------------------
>
> Key: CASSANDRA-18804
> URL: https://issues.apache.org/jira/browse/CASSANDRA-18804
> Project: Cassandra
> Issue Type: Bug
> Components: Accord
> Reporter: David Capwell
> Assignee: David Capwell
> Priority: Normal
> Labels: pull-request-available
> Fix For: 5.x
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> This work is forking CASSANDRA-18675 and pulling out the initial work and bug
> fixes, but does not actually include drop keyspace support, but does include
> the testing for add keyspaces.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
