Re: Tables become corrupted in Terminal Server Session by Mark
Mark
Tue Jul 08 02:17:44 CDT 2008
From experinace here using terminal server we find it best to have the data
on the same machine as the terminal server, as this cuts down to the minimum
the chance of a data corruption. We have also noticed since VFP9 that the
runtime, is far more sensitive to problems between the record count in the
table header and the one in the database container, where as before things
could continue to operate, they now tend to stop until someone sorts it out.
As for finding out why, we have spent a lot of time trying to do exactly
that. The main causes in our system seem to be file locks being left on
servers, our recreate index routine failing (again normal due to user rights
or files in use). The system being terminated either because of a crash, or
becasue someone just ended it rather than log of properly. After developing
an overnight reindex, and a timer based logout routine, we have improved
things, but we still have had to develope some routines to fix the data, and
seem to require them fairly regularly.
hope someone else on here has some bright idea's
"WP" <wpaul_nospam@arnet.com.ar> wrote in message
news:%23PQMTHL4IHA.3784@TK2MSFTNGP06.phx.gbl...
> My scenario:
> application's exe (vfp9) running from a file server, on 30 computers and
> there are also 10 sessions of terminal server who run on another server
> (not the file server). The users of the 10 sessions are 3 miles away from
> both file server and the terminal server. Connection is over a 2 MB point
> to point line, with a good ping of about 30 ms.
>
> Surprisingly, from time to time some tables on the file server become
> corrupted, and I can see that these tables had been in use by the "remote"
> people in terminal server mode. (These are the only people that can update
> the corrupted tables. The "local" people just open the tables with
> NOUPDATE).
>
> For example today, one of the tables that became corrupted, indicated with
> Foxfix5 a wrong record count. The table is small, about 15000 records,
> with a record length of 50, that means 750 kB.
>
> Is there a chance to get a corrupted table because of a micro cut in the
> connection to the remote office? I could understand this if it were a true
> client-server application, but it runs on a LAN and on terminal server,
> this means only video is updated. What else can be happening ?
>
>
> Walter
>