The dellog column family currently has a primary key that is a timestamp of the current datetime serialised with second resolution. Given that hundreds or thousands of records can be deleted every second, this makes it worse than useless as a backup strategy. Prepending the ALA Internal UUID to the primary key along with the timestamp will be enough to preserve records during bulk deletes and enable ordering of deletes for a given UUID to examine their contents.
The
dellog
column family currently has a primary key that is a timestamp of the current datetime serialised with second resolution. Given that hundreds or thousands of records can be deleted every second, this makes it worse than useless as a backup strategy. Prepending the ALA Internal UUID to the primary key along with the timestamp will be enough to preserve records during bulk deletes and enable ordering of deletes for a given UUID to examine their contents.https://github.com/AtlasOfLivingAustralia/biocache-store/blob/e0f58afac09cc03b50c1e2aef3f8f021cbd65056/src/main/scala/au/org/ala/biocache/dao/OccurrenceDAOImpl.scala#L1275