Seems to end up with an infinite loop of sorts once the last coverage of the NetCDF file is being purged (assuming there are multiple ones), with the NetCDF file re-populating its index as the granule is processed for deletion:
The purging code goes through the granule descriptor which in turn initializes the store, which I guess finds the database empty and re-initializes itself… we probably need to either avoid going through the granule descriptor (just to find the raster URL really) or have hints that avoid re-initialization of the NetCDF.
Also noticing, the code is assuming granule==file but for NetCDF one file maps to many granules, it should probably be processed once only.
Trying to delete all the granules related to a given NetCDF files, like this:
curl -XDELETE -u admin:geoserver "http://localhost:8080/geoserver/rest/workspaces/rm/coveragestores/hsf_ita/coverages/mycoverage/index/granules?filter=$filter&purge=metadata"
Seems to end up with an infinite loop of sorts once the last coverage of the NetCDF file is being purged (assuming there are multiple ones), with the NetCDF file re-populating its index as the granule is processed for deletion:
The purging code goes through the granule descriptor which in turn initializes the store, which I guess finds the database empty and re-initializes itself… we probably need to either avoid going through the granule descriptor (just to find the raster URL really) or have hints that avoid re-initialization of the NetCDF.
Also noticing, the code is assuming granule==file but for NetCDF one file maps to many granules, it should probably be processed once only.
As a further notice, the NetCDF index is stored in a postgis database, just like the mosaic index…. may not be relevant, but just in case.