cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Blake Eggleston (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-14058) Refactor read executor and response resolver, abstract read repair
Date Tue, 16 Jan 2018 22:56:00 GMT


Blake Eggleston commented on CASSANDRA-14058:

bq. Row/PartitionIteratorMergeListener has a dependency on BlockingReadRepair - guess this
should be ReadRepair instead? Or, if the R/PIML is intended to be BRR specific, we should
perhaps make them inner classes there?
They’re intended to be BlockingReadRepair specific, since the DigestReadRepair doesn’t
do any merging. Given their size though (RIML in particular), I’d rather not make them inner
classes. Maybe moving them into repair.blocking sub-package would be the way to go?
bq. Not a huge fan that ReadRepair has DigestResolver-specific methods - but I have no real
improvement suggestion here - either ReadRepair has DigestResolver-specific logic or DigestResolver
has read repair logic.
I think I intended that as sort of a shim that would go away after CASSANDRA-10726. Once the
strategy is configurable, there will probably be a factory class or something attached to
the table metadata that can handle that background repair case.

rebased on current trunk and pushed up with review fixes [here|]

> Refactor read executor and response resolver, abstract read repair
> ------------------------------------------------------------------
>                 Key: CASSANDRA-14058
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Blake Eggleston
>            Assignee: Blake Eggleston
>            Priority: Major
>             Fix For: 4.0
> CASSANDRA-10726 is stuck right now because the state of {{AbstractReadExecutor}} and
{{DataResolver}} make it difficult to cleanly implement. It also looks like some additional
read repair strategies might be added. This goal of this ticket is to clean up the structure
of some of the read path components to make CASSANDRA-10726 doable, and additional read repair
strategies possible.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message