Re: Size control of minot compaction

Posted by kunalkapoor on
URL: http://apache-carbondata-dev-mailing-list-archive.168.s1.nabble.com/Size-control-of-minor-compaction-tp103414p103485.html

Hi Zhangshunyu,
We should refactor the code and change the property name from "
carbon.major.compaction.size" to "carbon.compaction.size.threshold"( A
global property is exposed which defines the size after which segment would
not be considered for auto compaction). By doing this we can use the same
threshold for major and minor compaction. Let us avoid adding new property
for a minor compaction size threshold.

Minor compaction would consider the segments based on  "
carbon.compaction.size.threshold  " and "carbon.compaction.level.threshold".
Major would consider all segments with size below "
carbon.compaction.size.threshold".
Custom compaction should not consider any property and do a force
compaction(existing behaviour).

Thanks
Kunal Kapoor

On Tue, Nov 24, 2020 at 10:32 AM Kunal Kapoor <[hidden email]>
wrote:

> Hi Zhangshunyu,
> We should refactor the code and change the property name from "
> carbon.major.compaction.size" to "carbon.compaction.size.threshold"( A
> global property is exposed which defines the size after which segment would
> not be considered for auto compaction). By doing this we can use the same
> threshold for major and minor compaction. Let us avoid adding new property
> for a minor compaction size threshold.
>
> Consider 5 segments when carbon.compaction.threshold = 1GB:
>
> Minor compaction would consider the segments based on  "
> carbon.compaction.size.threshold  " and "carbon.compaction.level.threshold
> ".
> Major would consider all segments with size below "
> carbon.compaction.size.threshold".
> Custom compaction should not consider any property and do a force
> compaction(existing behaviour).
>
> Thanks
> Kunal Kapoor
>
> On Tue, Nov 24, 2020 at 7:32 AM Zhangshunyu <[hidden email]>
> wrote:
>
>> OK
>>
>>
>>
>> -----
>> My English name is Sunday
>> --
>> Sent from:
>> http://apache-carbondata-dev-mailing-list-archive.1130556.n5.nabble.com/
>>
>