problem if validation fails

[muhammadshoaib]

Good day,

I was trying to load the data into Ada, however, I had problem with validation in the first step. This made the whole system stopped from working and deleting the inapropriate study. I got the following error.

`Unable to provision, see the following errors:
1) Error injecting constructor, org.elasticsearch.client.ResponseException: PUT http://127.0.0.1:9200/data-covirna_lih.predicovid_dropped_fields: HTTP/1.1 400 Bad Request
{"error":{"root_cause":[{"type":"mapper_parsing_exception","reason":"Failed to parse mapping [data-covirna_lih.predicovid_dropped_fields]: name cannot be empty string"}],"type":"mapper_parsing_exception","reason":"Failed to parse mapping [data-covirna_lih.predicovid_dropped_fields]: name cannot be empty string","caused_by":{"type":"illegal_argument_exception","reason":"name cannot be empty string"}},"status":400}
  at org.ada.server.dataaccess.elastic.ElasticJsonCrudRepo.<init>(ElasticJsonCrudRepo.scala:23)
  while locating org.incal.core.dataaccess.AsyncCrudRepo<play.api.libs.json.JsObject, reactivemongo.bson.BSONObjectID> annotated with @com.google.inject.internal.UniqueAnnotations$Internal(value=1)
1 error`

[sherzinger]

confirmed

Related to special characters in column names/dataset names

[sherzinger]

@LBolzani when you have time please have a look at this. The issue is that a crash at the wrong time of the ETL process can cause a discrepancy between Mongo and ES, which results in a state that the UI cannot recover from. Direct DB intervention is necessary, so this should be prevented by any means.

Try to make iris.csv invalid by removing or editing fields and then load it. If you can't replicate let me know and I'll help.

[gh-osh]

@muhammadshoaib Was it a valid CSV?

[sherzinger]

Probably not. I’ve seen this error before and it can be triggered by certain format violations. If you have a file that triggers this problem (the DB corruption) please do share it with us in a secure way.

[peterbanda]

Guys, read the data set name: data-covirna_lih.predicovid_dropped_fields

@muhammadshoaib was most likely dropping some fields (i.e., running a transformation) and forgot to specify a field name (validation is probably missing there).

Btw. all special characters in field names are replaced by _, so that's definitely not a cause.