DBZ-7631 Change hashing strategy to use partition columns #126
Conversation
| @@ -665,4 +665,8 @@ private List<Object> getPartitionKeys(PartitionUpdate pu) { | |||
| return values; | |||
| } | |||
|
|
|||
| private int getPartitionQueueIndex(PartitionUpdate partitionUpdate) { | |||
There was a problem hiding this comment.
Can this method be extracted into a utility class in core module and re-used in all three modules?
|
@samssh Thank you for the PR. I left a comment related to a small refactoring to re-use the code. |
|
@jpechane You're welcome. Regarding the method refactoring, unfortunately, it's not possible to extract the method to a utility class in the core module because for every version of Cassandra we have a different artifact that contains the And regarding configuring the behavior, that's fine, and I will add a configuration property with an enum as soon as possible. |
|
@samssh I see, that's unfortunate. Let's forget about the refactoring then. |
samssh
force-pushed
the
DBZ-7631
branch
2 times, most recently
from
c39a0a5 to
534904b
Compare
…guration property. To ensure the proper ordering of events within the same partition, it's necessary to utilize the partition columns when determining the queue index for event insertion. To maintain the previous behavior, a property has been added to the config to allow the selection of the hashing strategy.
|
@jpechane It should be ready. |
|
@samssh Applied, thanks! Could you please create a PR against the core repo with docs update? |
|
Thank you. Of course! I'll get on it as soon as possible. |
To ensure the proper ordering of events within the same partition, it's necessary to utilize the partition columns when determining the queue index for event insertion.