The data analysis in elasticsearch cannot be queried through the api in the zipkin service

scene :

The zipkin link log normally generates link data in elasticsearch. When the data is queried through es, the traceId data exists, but the data before the current day's data cannot be queried by calling the zipkin api.

 The data of the day exists in es

 

 The data before the day also exists in es

 

The above shows that it is normal for zipkin to write data in es normally.

Question 1: Why can’t I call the zipkin interface to query the data before the day as follows

Where gisq_request_id=fe88132e-1190-4e36-b0b2-286bceb39df1 is the id before that day. 

 Note: The lookback parameter is the parameter carried by the api in zipkin. The default query is the current data lookback. The default is 86400000. If you want to query the previous data, you need 86400000*nday. Here I query the data of the previous 10 days

Where gisq_request_id=973fa4e7-91e2-4d30-82dd-ad6770144897 is the id of the day.

 This is very strange, the data exists in elasticsearch, why can't it be queried by calling zipkin's api? ? ? ? ?

problem analysis:

   Index information of api call exception in zipkin

 

API call normal index information in zipkin

Data Generation Structural Analysis

Under abnormal circumstances, the structure generated by the data contains the _q field, because this field is not excluded in the mapping template mapping

 

 Under normal data clearing, there is no _q field under _source, because this field is excluded in the mapping template mapping.

In addition, when zipkin/api/v2/traces?annotationQuery=gisq_request_id=fe88132e-1190-4e36-b0b2-286bceb39df1&lookback=864000000&limit=100 is called abnormally, the backend reports a null pointer.
The prompt is that json parsing failed

problem causes:

The zipkin index template may have been moved by others,

It may be to delete the index template through the es command

It may also be that the current zipkin index is deleted through the graphical interface, and the index data recreated in the code

Scene reproduction:

After creating the index through zipkin, then delete the index data of the day through the graphical interface

The shard of the regenerated index after deletion becomes 0, and the index created before deletion has 5 shards by default

Guess you like

Origin blog.csdn.net/qq_38423256/article/details/128900936