I have had several databases go corrupt over the weekend (fb 1.5.4 on fc8).

The circumstances are that there are several running C processes, some
connect to the database on startup and poll it, others connect, do
their thing and disconnect.
The hostname on these systems was changed (using the hostname command)
shortly before the corruption messages starting appearing in the
logfiles, mostly of the "wrong page type, expect x got y" variety,
where x can be 3 or 5 and so can y (but there may be others I didn't
notice). The times on the the isc_ files in the firebird directory
indicate that the ones named after the old hostname were still being
written after the one for the new name were being used as well.

Can anyone confirm that the hostname change is a probable cause or
suggest an alternative

(I know this wasn't the best idea, and in fact, I didn't do it but I
need to explain to the customer why/how it happened)