Re: GlobalDict should have 0 or 1 append dict but 2?

[ShaoFeng Shi <sh...@apache.org>]

Google this question and you will get the answer:
https://issues.apache.org/jira/browse/KYLIN-1540

2017-03-20 18:22 GMT+08:00 赵亮 <li...@fanli.com>:

> Thanks!
> I read the solution and I’ll try it .
> But there is still a question : how can I delete the last segments of a
> cube by using some rest-api ?
>
> 发件人: "446463844@qq.com" <44...@qq.com>
> 日期: 2017年3月20日 星期一 13:37
> 至: dev <de...@kylin.apache.org>, 赵亮 <li...@fanli.com>
> 主题: Re: GlobalDict should have 0 or 1 append dict but 2?
>
> solution:
> http://apache-kylin.74782.x6.nabble.com/java-lang-IllegalStateException-
> GlobalDict-dict-tableName-column-should-have-0-or-1-
> append-dict-but-y-td6848.html
> some steps:
> .1 find the different global_dictionary rowkey in segment
> .2 you should delete the segment first
> .3 delete the rowkey in Hbase tables which start with '/dict' and '/cube'
> ________________________________
> 446463844@qq.com
>
> From: 赵亮<ma...@fanli.com>
> Date: 2017-03-20 10:51
> To: dev@kylin.apache.org<ma...@kylin.apache.org>
> Subject: GlobalDict should have 0 or 1 append dict but 2?
> Hi all,
>        I build a cube with a column named “USER_ID” using global
> dictionary day by day. Last day I met a problem that this column has 2
> global dictionary like :
> [cid:image001.png@01D2A167.EE9792A0]
> In this pic , DW.INCR_D_USER_NEW_CHANNEL is the hive table name.
>
> I found that the dictionary ID of USER_ID in last segments is different
> from other segments:
> [cid:image002.png@01D2A167.EE9792A0]
>
>
> so my idea is delete the last segment and rebuild it , but I don’t know
> how to do it.
> I found there is an rest-api in https://issues.apache.org/
> jira/browse/KYLIN-1540 but is’t not in the kylin official doc. Is this
> api working ? if so , someone can give me a example？
>
> I also found a solution in http://apache-kylin.74782.x6.
> nabble.com/Re-more-than-1-append-dict-for-globalDict-td7028.html but it’s
> useless for me because I can not login cluster as Hadoop administrator,I
> just have the admin privilege of kylin.
> I wondering if there is a solution without using Hadoop cmd.
>
>
> In the above solution, I can’t not find how this error happened and I want
> know it.
>
> Thanks!
>
>
> Best wishes,
> Light
>



-- 
Best regards,

Shaofeng Shi 史少锋