Description
On 2021-09-29, we deployed changes to the data exporter and archive differ to automatically populate missingness columns to indicate deleted rows or typically-unavailable columns in the files we submit to delphi-epidata acquisition. From this day through 2021-10-04, acquisition improperly failed files that included the resulting missingness annotations.
This affected the following indicators:
- chng
- covid-act-now
- hhs
- hospital-admissions
- jhu-csse
- nchs-mortality
- usa-facts
This Kibana visualization shows how many files were improperly failed for each source:
We need to generate a patch for each of the affected indicators over these dates and upload it using the batch issue uploader. For everything except hospital-admissions, this can be done using the S3 ArchiveDiffer history. For hospital-admissions, we'll need to call in Maria.