incubator-couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chris Anderson <jch...@apache.org>
Subject Re: reduce_limit error
Date Mon, 04 May 2009 22:33:16 GMT
On Mon, May 4, 2009 at 3:23 PM, Adam Kocoloski <adam.kocoloski@gmail.com> wrote:
> Thanks for the note, Chris. It might be nice if we could figure out a way
> not to print "wazzup" N*1000 times when the test suite runs :-)  Other than
> that, cool!  Should make support a good bit easier.
>

Yes, that's an interesting side effect of Erlang's dump everything
crash reporting. I'm not sure what to do about that, but a more
general fix could be handy.

Maybe it'd be worth it in this particular case to convert to an
exit(normal) after sending the error message out to linked processes.

> Adam
>
> On May 4, 2009, at 6:08 PM, Chris Anderson wrote:
>
>> Devs,
>>
>> I'm checking in a patch that should cut down on the number of mailing
>> list questions asking why a particular reduce function is hella slow.
>> Essentially the patch throws an error if the reduce function return
>> value is not at least half the size of the values array that was
>> passed in. (The check is skipped if the size is below a fixed amount,
>> 200 bytes for now).
>>
>> I expect this heuristic will need fine tuning. Ideally we'd never
>> raise the error on "good" reduces, and always raise it on
>> "ill-conceived" ones. If you hit the error on a reduce that seems like
>> it should be considered good, please ping the list so we have an
>> opportunity to fine-tune.
>>
>> There is a config option:
>>
>> [query_server_config]
>> reduce_limit = true
>>
>> changing this to false will revert to the old behavior. Ideally very
>> few applications will require this config change, so if you find
>> yourself changing this setting, it's a sign you should mail the list.
>>
>> --
>> Chris Anderson
>> http://jchrisa.net
>> http://couch.io
>
>



-- 
Chris Anderson
http://jchrisa.net
http://couch.io

Mime
View raw message