Hello Tom,

>> TC> I've a 10.6GB FDB (made up of 4 long thin tables) and find that
>> TC> creating an index takes 3-4 hours+.
>> Can you show us the index information (using gstat)?

TC> Sure.
...
TC> Index CANDIDATECVWORDID (1)
TC> Depth: 3, leaf buckets: 42453, nodes: 110742402
TC> Average data length: 0.00, total dup: 90759213, max dup: 10356
^^^^^^^^ ^^^^^
That could cause the problem, I guess. You may try to make the index
more selective adding other field(s) to the index (to the end of its
definition): that increases the space used, but decreases CPU usage
when processing index nodes.

TC> Index CANDIDATENOTEWORDWORDID (1)
TC> Depth: 2, leaf buckets: 233, nodes: 602686
TC> Average data length: 0.00, total dup: 505470, max dup: 4060
You may want to change this index too.

As for low CPU usage in your case after 1/4th of restore/index
creation, and slow file growing - I have no ideas at the time: may be
OS plays a role here... Wait for others to offer different ideas.


--
Best regards,
Pavel Menshchikov
http://www.ls-software.com