Hi, all devs. Now Doris has supported random distribution for Duplicate Keys
table and Aggregate Keys table without replace(replace_if_not_null) type column
to solve data skew problem.
But now, It brings some problem when we do some speical queris on aggregate
table with random distribution.
Hi all,
The dev-1.0.1 is now closed. I'm just waiting for one last PR to resolve the
outer join bug.
--
此致！Best Regards
陈明雨 Mingyu Chen
Email:
morning...@apache.org
在 2022-06-24 10:06:05，"zhg yang" 写道：
>+1 we should maintain a version with bug fix but without new features
>Thanks
>Yang
nice work！
In the past few days, two or three users have raised the issue of data
skew. This solution is very good. I prefer the second of the two solutions,
so that users will not perceive it.
蔡聪辉 于2022年6月26日周日 21:49写道：
>
>
>
> Hi, all devs. Now Doris has supported random distribution for Dupli
It is now possible to detect the format of the tablet in the database at
that time through ADMIN SHOW TABLET STORAGE FORMAT, but there is no tools
to explicitly convert v1 to v2, we need to provide such a tool before
removing the code
Thanks
Yang Zhengguo
陈明雨 于2022年6月26日周日 11:12写道：
> Hi Guolei
The full update does not increase the complexity of the user's use. Under
normal circumstances, the user also needs to create a view, and the normal
view also needs to be calculated in time. Using a multi-table materialized
view is equivalent to caching some results. In order to reduce the user's
d
And pre-request feature may be we should improve the performance of insert
into select ，but this may be not pre， post may be also fine
Thanks
Yang Zhengguo
陈明雨 于2022年6月26日周日 11:55写道：
> I have read the DSIP, and I think the Instructions for use are clearer,
> but we need more detail designs.
I support solution 2. We should provide the same query interface for
different tables. The wrong query result is mainly caused by the query
planning not taking into account the new data shuffle method. This problem
is more like a bug than a feature, because I I think it's still a matter of
fixing t
Hi all,
The code branch for version 1.1 has been frozen.
Version 1.1 can be regarded as a bug fix and optimized version of v1.0, and
many new features that have been developed are not included in version 1.1.
And these functions are also the functions that users desperately need.
I recommend to
Hi Devs, I've update the DISP last weekend, if you are interest on this
feature, welcome to review and comment, thanks
https://cwiki.apache.org/confluence/display/DORIS/DSIP-018%3A+Support+Merge-On-Write+implementation+for+UNIQUE+KEY+data+model
Best
Chen Zhang
在 2022年6月24日 +0800 10:13，zhg yang ，写
I support solution2.
Although such queries with mismatched aggregation types are really
meaningless in real scenarios,
based on Doris's consistent design style,
we should allow users to submit such queries instead of banning them,
just like we now allow users to directly query ```select * from
agg_
1. Remove string type
Are we going to remove the string type in version 1.2?
If so, what will be the impact on users who have already used the string
type, and how to make it compatible
2. New Flink CDC Connector
Can it be released together with 1.1, this function has been tested in 1.1,
it should
For Agg table, data distribution way should keep the consistent semantics,
from this perspective, I think solution 2 is better,this reduces the cost
of user understanding.
I have another question that whether rewrite sql and add a shuffle may
cause performance drop in some scenes.
蔡聪辉 于2022年6月26
> I have another question that whether rewrite sql and add a shuffle may
cause performance drop in some scenes.
Yes, but that's the price random tables have to pay.
Random's agg table means that it can no longer perform complete
pre-aggregation operations.
So there will be a loss of code performan
I am willing to be the release manager.
张家峰 于2022年6月27日周一 12:31写道：
> 1. Remove string type
> Are we going to remove the string type in version 1.2?
> If so, what will be the impact on users who have already used the string
> type, and how to make it compatible
>
> 2. New Flink CDC Connector
> Ca
why remove string type ？
Thanks
Yang Zhengguo
陈明雨 于2022年6月27日周一 11:52写道：
> Hi all,
> The code branch for version 1.1 has been frozen.
> Version 1.1 can be regarded as a bug fix and optimized version of v1.0,
> and many new features that have been developed are not included in version
> 1.1.
>
15 matches
Mail list logo
