> > > Does some of them have
> > > > fixed cache setting ?
> > >
> > > I don't know what this is.
> >
> > Show us the output of gstat -h for each database, please
>
> Is this the information you would like to have ?
>
> Database header page information:
> Flags 0
> Checksum 12345
> Generation 13848809
> Page size 4096
> ODS version 10.1
> Oldest transaction 13823495
> Oldest active 13823496
> Oldest snapshot 13822165
> Next transaction 13848701
> Bumped transaction 1
> Sequence number 0
> Next attachment ID 0
> Implementation ID 16
> Shadow count 0
> Page buffers 0
> Next header page 0
> Database dialect 1
> Creation date Jun 14, 2006 13:37:55
> Attributes force write
>
> Variable header data:
> Sweep interval: 20000
> *END*
>

25,000 between the oldest active and the next.
approx 76,000 transactions per day based on total numbers since June 14

it's a large gap but at that rate of transactions, it doesn't look too
silly - about 8 hours behind.
really round numbers here, but at your user level (maybe average 75) that's
about 1000 transaction per user per day (about 2 trans per minute). There's
no way of telling if one user is keeping it back 25 days behind the rest or
all users are keeping it 8 hours behind the leader.
It would be interesting to see if this gap closes during quieter times and
widens again during busy times or if it remains that wide all the time.
Maybe my logic is warped....
Alan