I agree that there's no bug here. But there are some simple things that
could be done (at least under UNIX) to notice & flag potential disasters.
E.g., the original log file could be periodically re-opened (readonly) and
a check made to see that it has the same device & inode as when it was
originally opened. If not, do something.
Or, probably better, on startup a hard link is made to the log file (in a
private zope-specific directory) and that link is opened for logging. On
shutdown the hard link is removed unless its link count has dropped to 1,
in which case it is not removed and some admin alert action is triggered.
Such a safeguard could be disabled by experts in the config.
For more information about ZODB, see the ZODB Wiki:
ZODB-Dev mailing list - ZODB-Dev@zope.org