Getting to know Cassandra internals

[Chris Vest <mr...@gmail.com>]

Hi,

I'm attacking https://issues.apache.org/jira/browse/CASSANDRA-5351 in order
to familiarise myself with the code.

I've added a repairedAt field to SSTableMetadata as suggested, and to its
non-legacy serialised form. Now I'm trying to figure out where the field
should be used.
My guess is, that if I change getScanners(sstables, range) in Abstract- and
LeveledCompactionStrategy to not include SSTables that have repairedAt
stamps in their metadata, in the returned list of scanners, then that would
solve the part about not generating MerkleTrees for already repaired
sstables. Then the next step will be writing repairedAt metadata for tables
that do get repaired.

Is this thinking correct?

Cheers,
Chris

Re: Getting to know Cassandra internals

[Jonathan Ellis <jb...@gmail.com>]

You could probably get that to work but I think it will be more
straightforward to restrict the set of sstables being validated before
calling getScanners.

On Sun, Mar 17, 2013 at 6:05 AM, Chris Vest <mr...@gmail.com> wrote:
> Hi,
>
> I'm attacking https://issues.apache.org/jira/browse/CASSANDRA-5351 in order
> to familiarise myself with the code.
>
> I've added a repairedAt field to SSTableMetadata as suggested, and to its
> non-legacy serialised form. Now I'm trying to figure out where the field
> should be used.
> My guess is, that if I change getScanners(sstables, range) in Abstract- and
> LeveledCompactionStrategy to not include SSTables that have repairedAt
> stamps in their metadata, in the returned list of scanners, then that would
> solve the part about not generating MerkleTrees for already repaired
> sstables. Then the next step will be writing repairedAt metadata for tables
> that do get repaired.
>
> Is this thinking correct?
>
> Cheers,
> Chris



-- 
Jonathan Ellis
Project Chair, Apache Cassandra
co-founder, http://www.datastax.com
@spyced