A bit of background :

We use a number of laptops on remote screening units to allow information to
be directly entered at whatever location the unit is at.  On a daily basis
an extract is performed from the main server system to removable flash ATA
drives which are subsequently sent to the mobile location.  On return these
updated disks are imported to the server.

This is performed by having a cache database residing on the removable disk
(100Mb Cache.Dat).  At the server side this is either connected using ECP or
by programatically mounting/dismounting the database as required (server is
configured with a local namespace which points to the removable drive).
Transfer is done using the merge command ie M ^local=^|remote|local.

We have a number of intermittent problems where the disk appears to be
created ok but then generates windows BOOTEX logs when loaded at laptop or
back at server which results in corruption of the Cache database.  My guess
is that this may be a caching issue which leaves windows thinking that the
file/volume was still in use etc

Ok, questions :

1.  Does a call to ^DISMOUNT ensure buffers are flushed before database is
actually dismounted.  If not is there a way to ensure this ($ZU(51)
perhaps).
2.  Is this scenario sensible ie having a Cache database residing on a
removable disk which is moved between more than one Cache system ?
3.  Any miscellaneous settings that may ensure all writes to this database
are updated before unmounting/disconnecting

General system settings are Win2k, Cache Versions 5.0.2 (5.0.5 on one
server).  Removable disks are formatted as FAT, each holding a 100Mb Cache
database.  Each disk is built using the same 100Mb Database file.

Any ideas much appreciated.  Also if anybody has similar applications I'd be
interested in how you update / synchronise the main database.  I've seen
some info on the new 5.1 object syncronisation but at the moment this
application is straightforward COS.

Cheers,

Mike




Reply via email to