Web index generation should probably be based on jsonfile.log
Our current gen-indexes.sh is a silly pile of find(1) commands looking for build logs in a large directory hierarchy. I think all the information it needs in in /srv/debomatic/jsonfile.log, so it should probably just slurp that in and base the indices on it.
The only thing I'm wondering about this is logfile rotation, which we're not doing at the moment so there are 146k lines spanning two years. Maybe do that every new year or something and accept that the indices get cleared at that point?