Closed felix-geovelo closed 3 months ago
Indexing by rank is still needed for a reverse-only database. It calcuzwhich streets belong to which city, which city to which state etc. Only the search index creation is skipped. That’s maybe 10% of total import time.
Thanks for the quick and clear reply, maybe it could be a useful add to the doc saying that --reverse-only save up to 10% of import time on behalf of 5% disk space.
Thanks again
Describe the bug I can't see any difference in the import process with and without --reverse-only. In both case I have the same logs output.
To Reproduce I tried to import a small PBF file with and without the --reverse-only param to see the difference in the import logs. But there is apparently none. I was expecting that with --reverse-only, all logs looking like "Rank ... ETA" would disappear but that was not the case (i.e that all indexing stage would be skipped) Does this feature still works ? Is it supposed to save a lot of time during the import process for large PBF file ?
Software Environment (please complete the following information):
Hardware Configuration (please complete the following information):
Postgresql Configuration:
RAS
Nominatim Configuration:
Variable set in .env : NOMINATIM_DATABASE_DSN & NOMINATIM_DATABASE_WEBUSER
More info:
I'm trying to create a nominatim DB to import into photon after, so --reverse-only could help me to save some time during import process by skipping indexing stages