Hi Lisa,

We have regular backups to an external HD (daily and weekly in 2 day and 2 week rotation sets, with occasional additional manual backups taken offsite), but due to the way that the database corruption "snuck up" on us (it occurred gradually over a number of weeks, rather than as a sudden and complete failure), the corrupt data was present to various extents in all of the backups in the regular rotation.

While it may be possible to "merge" the various datasets to get some data back, there have been upgrades to the software between some of them resulting in slightly different db schemas. Because of the large number of man hours that would be required to attempt such a recovery, and because it is certain we still wouldn't get 100% of the data back, we're forced to live with some data loss of both posts and user accounts.

As a result of this event, we are increasing the backup schedule dramatically, to a 7 day daily rotation, a 4 week weekly rotation, and a 3 month monthly rotation. This still won't guarantee no future data loss (that would require more than one server and a different database technology), but it will mean a good backup should be "closer" than several months back in time.