My client had a customer that was experiencing this problem, on a Novell server running version 3.26. The data resided on the server; the application was written in VFP 3.0 and resides on the client. There server would crash almost weekly, invariably corrupting at least one table.

Does anyone know if the FLUSH command forces Novell to commit its buffered file data to disk, or does it merely commit the VFP buffers to Novell? Would it be wise to use the Novell library routines (the product name escapes me now), and force a call to the server periodically, say in the following code:

END TRANSACTION
IF TXNLEVEL() = 0
*-- Commit Novell buffers to the hard disk.
ENDIF


Ed


>We have an application that stores data on a Novell server.
>The server was shutdown while the application was still
>running. Several new records were lost. (must have been
>in memory when shutdown occured ??) I know foxpro sent the
>records because a label is printed after the inserts take
>place.
>
>When the server came back up the application had a "not
>a table" error message on 2 tables. Used desalvage to fix but
>there were several records in each that were missing.
>
>
>Is it likely that the records had not been written and
>were in memory?
>
>Any ideas on how to prevent this from happening again?
>
>Thanks,
>Dave