[ 
https://issues.apache.org/jira/browse/KYLIN-2945?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Yifei Wu updated KYLIN-2945:
----------------------------
    Description: 
Currently, absolute path is used in the global dictionary for saving the 
location of specific info like the next json.
{code}
{
  "uuid" : "b82505d8-5b40-4009-8839-8456500ea6a8",
  "last_modified" : 0,
  "version" : "2.3.0.20500",
  "source_table" : "KYLIN_YIFEI_1011_TESTGLOBALDICTMIGRATION.KYLIN_SALES",
  "source_column" : "PRICE",
  "source_column_index" : 6,
  "data_type" : "decimal(19,4)",
  "input" : {
    "path" : 
"hdfs://sandbox.hortonworks.com:8020/kylin/kylin_default/kylin-d213a963-b5e3-4e75-a1a2-8af93ca83a80/global_dict_cube/fact_distinct_columns/KYLIN_SALES.PRICE",
    "size" : 81150,
    "last_modified_time" : 1508133611074
  },
  "dictionary_class" : "org.apache.kylin.dict.AppendTrieDictionary",
  "cardinality" : 8878
}^@hdfs://master:8020/kylin/kylin_yifei_1011_testGlobalDictMigration/resources/GlobalDict/dict/KYLIN_YIFEI_1011_TESTGLOBALDICTMIGRATION.KYLIN_SALES/PRICE/
{code}
But if you change the path configuration or migrate the data to another cluster 
which node name is different or original host cannot be reached, you will meet 
problems in reusing it, like rebuilding and merge dictionary or transfer column 
from dictionary in building cube, for the dictionary info cannot be found.

  was:
Currently, absolute path is used in the global dictionary for saving the 
location of specific info. But if you change the host configuration or migrate 
the data to another data. So you may meet problems that the dictionary cannot 
be found and reuse it after configuration changed or data migration.
{code}
{
  "uuid" : "b82505d8-5b40-4009-8839-8456500ea6a8",
  "last_modified" : 0,
  "version" : "2.3.0.20500",
  "source_table" : "KYLIN_YIFEI_1011_TESTGLOBALDICTMIGRATION.KYLIN_SALES",
  "source_column" : "PRICE",
  "source_column_index" : 6,
  "data_type" : "decimal(19,4)",
  "input" : {
    "path" : 
"hdfs://sandbox.hortonworks.com:8020/kylin/kylin_default/kylin-d213a963-b5e3-4e75-a1a2-8af93ca83a80/global_dict_cube/fact_distinct_columns/KYLIN_SALES.PRICE",
    "size" : 81150,
    "last_modified_time" : 1508133611074
  },
  "dictionary_class" : "org.apache.kylin.dict.AppendTrieDictionary",
  "cardinality" : 8878
}^@hdfs://master:8020/kylin/kylin_yifei_1011_testGlobalDictMigration/resources/GlobalDict/dict/KYLIN_YIFEI_1011_TESTGLOBALDICTMIGRATION.KYLIN_SALES/PRICE/
{code}



> global dict specific info use absolute path when saved it
> ---------------------------------------------------------
>
>                 Key: KYLIN-2945
>                 URL: https://issues.apache.org/jira/browse/KYLIN-2945
>             Project: Kylin
>          Issue Type: Improvement
>            Reporter: Yifei Wu
>
> Currently, absolute path is used in the global dictionary for saving the 
> location of specific info like the next json.
> {code}
> {
>   "uuid" : "b82505d8-5b40-4009-8839-8456500ea6a8",
>   "last_modified" : 0,
>   "version" : "2.3.0.20500",
>   "source_table" : "KYLIN_YIFEI_1011_TESTGLOBALDICTMIGRATION.KYLIN_SALES",
>   "source_column" : "PRICE",
>   "source_column_index" : 6,
>   "data_type" : "decimal(19,4)",
>   "input" : {
>     "path" : 
> "hdfs://sandbox.hortonworks.com:8020/kylin/kylin_default/kylin-d213a963-b5e3-4e75-a1a2-8af93ca83a80/global_dict_cube/fact_distinct_columns/KYLIN_SALES.PRICE",
>     "size" : 81150,
>     "last_modified_time" : 1508133611074
>   },
>   "dictionary_class" : "org.apache.kylin.dict.AppendTrieDictionary",
>   "cardinality" : 8878
> }^@hdfs://master:8020/kylin/kylin_yifei_1011_testGlobalDictMigration/resources/GlobalDict/dict/KYLIN_YIFEI_1011_TESTGLOBALDICTMIGRATION.KYLIN_SALES/PRICE/
> {code}
> But if you change the path configuration or migrate the data to another 
> cluster which node name is different or original host cannot be reached, you 
> will meet problems in reusing it, like rebuilding and merge dictionary or 
> transfer column from dictionary in building cube, for the dictionary info 
> cannot be found.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to