[ 
https://issues.apache.org/jira/browse/CASSANDRA-8833?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14328025#comment-14328025
 ] 

Benedict commented on CASSANDRA-8833:
-------------------------------------

bq. there may be unintended side-effects to some of those changes

I have no doubt, which is partially why I consider this premature. We had a 
period of problems (not isolated to this, I want to reiterate), and we've had a 
very recent attempt to put them to bed. But not let that settle down to see how 
successful it has been. I'm expecting a few more minor hiccups. If we see some 
more major ones, especially regressions, that's a different matter.

bq. Thus far we've avoided platform-specific code-paths as much as possible, 
but that seems a simple enough solution that it would be worth looking into.

I'd actually be totally cool with (and maybe even prefer) making this a 
universal behaviour. That way we don't have multiple mappings of the same file, 
and we still retain most of the upside, with no behavioural split.

> Stop opening compaction results early
> -------------------------------------
>
>                 Key: CASSANDRA-8833
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8833
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Marcus Eriksson
>             Fix For: 2.1.4
>
>
> We should simplify the code base by not doing early opening of compaction 
> results. It makes it very hard to reason about sstable life cycles since they 
> can be in many different states, "opened early", "starts moved", "shadowed", 
> "final", instead of as before, basically just one (tmp files are not really 
> 'live' yet so I don't count those). The ref counting of shared resources 
> between sstables in these different states is also hard to reason about. This 
> has caused quite a few issues since we released 2.1
> I think it all boils down to a performance vs code complexity issue, is 
> opening compaction results early really 'worth it' wrt the performance gain? 
> The results in CASSANDRA-6916 sure look like the benefits are big enough, but 
> the difference should not be as big for people on SSDs (which most people who 
> care about latencies are)
> WDYT [~benedict] [~jbellis] [~iamaleksey] [~JoshuaMcKenzie]?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to