[
https://issues.apache.org/jira/browse/CASSANDRA-4180?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13619595#comment-13619595
]
Jason Brown commented on CASSANDRA-4180:
----------------------------------------
On the whole, LGTM.
I think Jonathan and I had the same basic notion of how to implement this, and
we ran into many of the same problems (schema loading borked hard and fast).
Admittedly, the one-off sstable reads were what kept kicking me in the shins
(repeatedly), and what kept taking me down wrong paths. I agree about the fishy
smell with knowledge of the SSTable format spread around, but minimally that's
for a another ticket.
We can probably remove the {code}output != null{code} check in
ColumnIndex.Builder.add() as LCR is no longer passing in null (on the first
pass).
Jonathan mentioned he didn't try to scrub, but I just tried it out locally, and
it failed with an OOM error. I'd look into it more, but brain power running out
this late at night. I'm attaching a file with the scrub errors. Will look into
it tomorrow.
> Single-pass compaction for LCR
> ------------------------------
>
> Key: CASSANDRA-4180
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4180
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Sylvain Lebresne
> Assignee: Jonathan Ellis
> Labels: compaction
> Fix For: 2.0
>
> Attachments: scrub-error.txt
>
>
> LazilyCompactedRow reads all data twice to compact a row which is obviously
> inefficient. The main reason we do that is to compute the row header.
> However, CASSANDRA-2319 have removed the main part of that row header. What
> remains is the size in bytes and the number of columns, but it should be
> relatively simple to remove those, which would then remove the need for the
> two-phase compaction.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
