Metadata inconsistency between coordinator query and metadata segment query

Hi Guys,

I working on a setup mainly doing batch ingestion (pstn cdr).

I do have an inconsistency while trying to get metadata of my ptsn_cdr datasource :

  • While running an REST call like this : GET /druid/coordinator/v1/metadata/datasources?full

I do get all my column (25 column) and metrics (6 metrics) :

“dimensions”: “account,direction,time,switch_ip,call_type,numb_plan,calling_cat,nat_call_num,calling_num,nat_add_part,add_call_pat_ad,type_called_num,num_plan_called,called_num,cat_real_num,typ_real_num,nat_real_called,real_called,billing_mode,service_code,release_loc,operator_id,circuit_id,in_trunk_group,out_trunk_group”,

“metrics”: “count,connect_duration,connect_ringing,total_duration,max_duration,min_duration”,

  • While running an REST call like this : POST /druid/v2 with the following JSON query:

{

“merge”: false,

“analysisTypes”:[“cardinality”],

“queryType”:“segmentMetadata”,

“dataSource”:“pstn_cdr”,

“intervals”:[“2017-04-24T23:59:00+00:00/2017-05-01T23:59:00+00:00”]

}

I do get just limited view on my column (only 19) and no metric at all :

[

{

“id”: “pstn_cdr_2017-04-24T00:00:00.000Z_2017-04-25T00:00:00.000Z_2017-05-18T17:33:06.453Z”,

“intervals”: null,

“columns”:
{

“__time”:
{

}

Do you get anything different if you do “merge”: true?

Hi Gian,

Thanks for your help.

After more research and debugging, I found out that my SW that generate the source json file was suffering from few bugs: column name was getting wrong.

(I was not able to identify it so easily as I do have millions of row’s in hundreds of files :slight_smile: !)

I do believe that this mis-alignement between column name and json key was creating the issue.

There is still something weird while using Superset tool as UI: metrics are not taken as metrics but as normal dimension …

I will investigate this one now !

Sorry for posting a not relevant message to the list.

Thanks Matt.