[
https://issues.apache.org/jira/browse/CASSANDRA-19325?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yifan Cai updated CASSANDRA-19325:
----------------------------------
Bug Category: Parent values: Code(13163) (was: Parent values:
Correctness(12982)Level 1 values: Recoverable Corruption / Loss(12986))
Description:
The ranges used in the analytics library do not have the consistent notation
with Cassandra. The analytics library, as in the cassandra ecosystem, should
use the open-closed range notation consistently, to avoid potential bugs in
implementation.
Besides that, during write process, the split sub-ranges are unordered. It does
not seem to affect correctness, but can be confusing.
was:
The range splitting implementation can produce the following false results.
- Given a tiny range, it can produce duplicated ranges, leading to spark
executors working on the same data set.
- The produced ranges are closed on both ends, making the same tokens being
shared by 2 ranges, leading to data duplication.
Besides the splitting error, during write process, the split sub-ranges are
unordered. It does not seem to affect correctness, but can be confusing.
Summary: [Analytics] Fix range split and use open-closed range
notation consistently (was: [Analytics] Fix range splitting that can produce
overlapping ranges)
> [Analytics] Fix range split and use open-closed range notation consistently
> ---------------------------------------------------------------------------
>
> Key: CASSANDRA-19325
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19325
> Project: Cassandra
> Issue Type: Bug
> Components: Analytics Library
> Reporter: Yifan Cai
> Assignee: Yifan Cai
> Priority: Normal
> Time Spent: 2h 50m
> Remaining Estimate: 0h
>
> The ranges used in the analytics library do not have the consistent notation
> with Cassandra. The analytics library, as in the cassandra ecosystem, should
> use the open-closed range notation consistently, to avoid potential bugs in
> implementation.
> Besides that, during write process, the split sub-ranges are unordered. It
> does not seem to affect correctness, but can be confusing.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
